CodeCanyon

Elon Musk’s AI company, xAI, has missed a self-imposed deadline to publish a finalized AI safety framework, as noted by watchdog group The Midas Project. xAI isn’t exactly known for its strong commitments to AI safety as it’s commonly understood. A recent report found that the company’s AI chatbot, Grok, would undress photos of women when asked. Grok can also be considerably more crass than chatbots like Gemini and ChatGPT, cursing without much restraint to speak of. Nonetheless, in February at the AI Seoul Summit, a global gathering of AI leaders and stakeholders, xAI published a draft framework outlining the company’s approach to AI safety. The eight-page document laid out xAI’s safety priorities and philosophy, including the company’s benchmarking protocols and AI model deployment considerations. As The Midas Project noted in a blog post on Tuesday, however, the draft only applied to unspecified future AI models “not currently in development.” Moreover, it failed to articulate how xAI would identify and implement risk mitigations, a core component of a document the company signed at the AI Seoul Summit. In the draft, xAI said that it planned to release a revised version of its safety policy “within three months” — by May 10. The deadline came and went without acknowledgement on xAI’s official channels. Despite Musk’s frequent warnings of the dangers of AI gone unchecked, xAI has a poor AI safety track record. A recent study by SaferAI, a nonprofit aiming to improve the accountability of AI labs, found that xAI ranks poorly among its peers, owing to its “very weak” risk management practices. That’s not to suggest other AI labs are faring dramatically better. In recent months, xAI rivals including Google and OpenAI have rushed safety testing and have been slow to publish model safety reports (or skipped publishing reports altogether). Some experts have expressed concern that the seeming deprioritization of safety efforts is coming at a time when AI is more capable — and thus potentially dangerous — than ever. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW

Work-Bench, whose portfolio includes unicorns Spring Health and Socure, just raised a $160 million Fund IV to quadruple down on backing companies in New York City. The firm announced the raise in a blog post on Monday, saying the latest fund will support seed-stage founders building enterprise software. Checks will range from $2 million to $4 million, and the team is looking specifically for companies building in AI, cybersecurity, developer tools, and enterprise applications. Jonathan Lehr, the firm’s co-founder, told TechCrunch that the company hopes to invest in 23-25 companies out of Fund IV. He said that this was the smoothest fundraise yet for his firm, owing it to the fact that Work-Bench has stayed “disciplined” since launching in 2013, “which matters more than ever in today’s AI frenzied environment.”

Yuga Labs, known as the creator of Bored Ape Yacht Club and other notable NFTs, sold its iconic CryptoPunks to the nonprofit NODE Foundation, which focuses on preserving digital art. The deal details are undisclosed, and a representative declined to comment. Back in April, NODE announced it received a $25 million grant to “build the future of digital art” from investor Micky Malka and Becky Kleiner. The former is chair of the foundation. Yuga Labs purchased CryptoPunks back in 2022 from Larva Labs amid the digital token craze that saw NFTs sell for millions of dollars. The highest-priced CryptoPunk NFT sold for nearly $24 million worth of crypto, according to the CryptoPunks website. The markets have slowed since then, however. Malka said that CryptoPunks “sparked a cultural movement,” and that NODE will “future-proof this landmark work,” to make it more accessible for people to engage with.

Popular legal AI tool Harvey will now be using leading foundation models from Anthropic and Google, moving beyond strictly using OpenAI’s, Harvey announced in a blog post on Tuesday. This is noteworthy because Harvey is one of the OpenAI Startup Fund’s most successful early-backed portfolio companies. The OpenAI Startup Fund is an OpenAI-associated fund to back companies developing products on top of AI technologies, primarily OpenAI’s own. While Harvey says it’s not abandoning OpenAI, merely adding more models and clouds, this is still a huge coup for OpenAI’s big competitors. Harvey is one of the first four startups that the OpenAI Startup Fund backed, it said in December 2022. That was back when OpenAI CEO Sam Altman was still running the fund. (The others in that first cohort include Descript, Mem, and Speak.) Harvey, which has grown like mad since then, is now a $3-billion-valuation startup, it said in February, when it announced a $300 million Series D led by Sequoia, with other big names like Coatue, Kleiner Perkins, and the OpenAI Fund piling in. Interestingly, Google’s venture arm, GV, led Harvey’s $100 million Series C in July 2024 (and the OpenAI Fund participated in that round, too). But Harvey didn’t immediately adopt Google’s AI models after it put Google’s corporate venture firm on its cap table. (GV participated in Harvey’s Series D as well.) So, what convinced Harvey to move beyond OpenAI’s models now? The startup’s internally developed benchmark, dubbed BigLaw, showed that a wide variety of foundation models are growing increasingly adept at a range of legal tasks and some are better at specific tasks than others. Instead of spending its efforts training models, Harvey figured, it could simply embrace high-performing, reasoning foundation models from other vendors (e.g. Google and Anthropic via Amazon’s cloud) and then fine-tune them for the legal market. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW Using a variety of models will also help as Harvey creates AI agents, the company says. “In less than a year, seven models (including three non-OAI models) now outperform the originally benchmarked Harvey system on BigLaw Bench,” Harvey wrote in the blog post. Harvey’s benchmark also showed that different foundation models are better at specific legal tasks than others. For instance, it says Google’s Gemini 2.5 Pro “excels” at legal drafting but “struggles” with pre-trial tasks like writing oral arguments because the model doesn’t fully understand “complex evidentiary rules like hearsay.” OpenAI’s o3 does such pre-trial tasks well, according to Harvey’s testing, with Anthropic’s Claude 3.7 Sonnet following close behind. Results from Harvey’s internal benchmarking.Image Credits:Harvey In its blog post, Harvey says that it will also now join the growing ranks of those sharing a public leaderboard of model benchmark performance. Its board will rank how major reasoning models are doing on legal tasks. And the company won’t just boil rankings down to a single number, but will also publish research where “top lawyers provide nuanced insights into model performance not captured by single-score benchmarks.” So, not only is OpenAI-backed Harvey adopting competitors’ models, it is also amping up the pressure on its backers (including Google) to keep proving themselves. Not that OpenAI should worry much on that score. While AI benchmarking is growing increasingly complex and somewhat political, this is a world where OpenAI still shines. “We are incredibly fortunate to have OpenAI as an investor in Harvey and key collaborator in our product,” Harvey CEO Winston Weinberg told TechCrunch in a statement. “And, we are energized to add to our options for customers as we continue to serve the needs of our customers globally.”

On Tuesday, WhatsApp scored a major victory against NSO Group when a jury ordered the infamous spyware maker to pay more than $167 million in damages to the Meta-owned company. The ruling concluded a legal battle spanning more than five years, which started in October 2019 when WhatsApp accused NSO Group of hacking more than 1,400 of its users by taking advantage of a vulnerability in the chat app’s audio-calling functionality. The verdict came after a week-long jury trial that featured several testimonies, including NSO Group’s CEO Yaron Shohat and WhatsApp employees who responded and investigated the incident. Even before the trial began, the case had unearthed several revelations, including that NSO Group had cut off 10 of its government customers for abusing its Pegasus spyware, the locations of 1,223 of the victims of the spyware campaign, and the names of three of the spyware maker’s customers: Mexico, Saudi Arabia, and Uzbekistan. TechCrunch read the transcripts of the trial’s hearings and is highlighting the most interesting facts and revelations that came out. We will update this post as we learn more from the cache of more than 1,000 pages. Testimony described how the WhatsApp attack worked The zero-click attack, which means the spyware required no interaction from the target, “worked by placing a fake WhatsApp phone call to the target,” as WhatsApp’s lawyer Antonio Perez said during the trial. The lawyer explained that NSO Group had built what it called the “WhatsApp Installation Server,” a special machine designed to send malicious messages across WhatsApp’s infrastructure mimicking real messages. “Once received, those messages would trigger the user’s phone to reach out to a third server and download the Pegasus spyware. The only thing they needed to make this happen was the phone number,” said Perez. NSO Group’s research and development vice president Tamir Gazneli testified that “any zero-click solution whatsoever is a significant milestone for Pegasus.” NSO Group confirms it targeted an American phone number as a test for the FBI Contact Us Do you have more information about NSO Group, or other spyware companies? From a non-work device and network, you can contact Lorenzo Franceschi-Bicchierai securely on Signal at +1 917 257 1382, or via Telegram and Keybase @lorenzofb, or email. For years, NSO Group has claimed that its spyware cannot be used against American phone numbers, meaning any cell number that starts with the +1 country code. In 2022, The New York Times first reported that the company did “attack” a U.S. phone but it was part of a test for the FBI. NSO Group’s lawyer Joe Akrotirianakis confirmed this, saying the “single exception” to Pegasus not being able to target +1 numbers “was a specially configured version of Pegasus to be used in demonstration to potential U.S. government customers.” The FBI reportedly chose not to deploy Pegasus following its test. How NSO Group’s government customers use Pegasus NSO’s CEO Shohat explained that Pegasus’ user interface for its government customers does not provide an option to choose which hacking method or technique to use against the targets they are interested in, “because customers don’t care which vector they use, as long as they get the intelligence they need.” In other words, it’s the Pegasus system in the backend that picks out which hacking technology, known as an exploit, to use each time the spyware targets an individual. NSO Group’s headquarters shares the same building as Apple In a funny coincidence, NSO Group’s headquarters in Herzliya, a suburb of Tel Aviv in Israel, is in the same building as Apple, whose iPhone customers are also frequently targeted by NSO’s Pegasus spyware. Shohat said NSO occupies the top five floors and Apple occupies the remainder of the 14-floor building. “We share the same elevator when we go up,” Shohat said during testimony. The fact that NSO Group’s headquarters are openly advertised is somewhat interesting on its own. Other companies that develop spyware or zero-days like the Barcelona-based Variston, which shuttered in February, was located in a co-working space while claiming on its official website to be located somewhere else. NSO Group admitted that it kept targeting WhatsApp users after the lawsuit was filed Following the spyware attack, WhatsApp filed its lawsuit against NSO Group in November 2019. Despite the active legal challenge, the spyware maker kept targeting the chat app’s users, according to NSO Group’s research and development vice president Tamir Gazneli. Gazneli said that “Erised,” the codename for one of the versions of the WhatsApp zero-click vector, was in use from late-2019 up to May 2020. The other versions were called “Eden” and “Heaven,” and the three were collectively known as “Hummingbird.” NSO says it employs hundreds of people NSO Group’s CEO Yaron Shohat disclosed a small but notable detail: NSO Group and its parent company, Q Cyber, have a combined number of employees totalling between 350 and 380. Around 50 of these employees work for Q Cyber. NSO Group describes dire finances During the trial, Shohat answered questions about the company’s finances, some of which were disclosed in depositions ahead of the trial. These details were brought up in connection with how much in damages the spyware maker should pay to WhatsApp. According to Shohat and documents provided by NSO Group, the spyware maker lost $9 million in 2023 and $12 million in 2024. The company also revealed it had $8.8 million in its bank account as of 2023, and $5.1 million in the bank as of 2024. Nowadays, the company burns through around $10 million each month, mostly to cover the salaries of its employees. Also, it was revealed that Q Cyber had around $3.2 million in the bank both in 2023 and 2024. During the trial, NSO revealed its research and development unit — responsible for finding vulnerabilities in software and figuring out how to exploit them — spent some $52 million in expenses during 2023, and $59 million in 2024. Shohat also said that NSO Group’s customers pay “somewhere in the range” between $3 million and “ten times that” for access to its Pegasus spyware. Factoring in these numbers, the spyware maker was hoping to get away with paying little or no damages. “To be honest, I don’t think we’re able to pay anything. We are struggling to keep our head above water,” Shohat said during his testimony. “We’re committing to my [chief financial officer] just to prioritize expenses and to make sure that we have enough money to meet our commitments, and obviously on a weekly basis.” First published on May 10, 2025 and updated with additional details.

On Tuesday, WhatsApp scored a major victory against NSO Group when a jury ordered the infamous spyware maker to pay more than $167 million in damages to the Meta-owned company. The ruling concluded a legal battle spanning more than five years, which started in October 2019 when WhatsApp accused NSO Group of hacking more than 1,400 of its users by taking advantage of a vulnerability in the chat app’s audio-calling functionality. The verdict came after a week-long jury trial that featured several testimonies, including NSO Group’s CEO Yaron Shohat and WhatsApp employees who responded and investigated the incident. Even before the trial began, the case had unearthed several revelations, including that NSO Group had cut off 10 of its government customers for abusing its Pegasus spyware, the locations of 1,223 of the victims of the spyware campaign, and the names of three of the spyware maker’s customers: Mexico, Saudi Arabia, and Uzbekistan. TechCrunch read the transcripts of the trial’s hearings and is highlighting the most interesting facts and revelations that came out. We will update this post as we learn more from the cache of more than 1,000 pages. Testimony described how the WhatsApp attack worked The zero-click attack, which means the spyware required no interaction from the target, “worked by placing a fake WhatsApp phone call to the target,” as WhatsApp’s lawyer Antonio Perez said during the trial. The lawyer explained that NSO Group had built what it called the “WhatsApp Installation Server,” a special machine designed to send malicious messages across WhatsApp’s infrastructure mimicking real messages. “Once received, those messages would trigger the user’s phone to reach out to a third server and download the Pegasus spyware. The only thing they needed to make this happen was the phone number,” said Perez. NSO Group’s research and development vice president Tamir Gazneli testified that “any zero-click solution whatsoever is a significant milestone for Pegasus.” NSO Group confirms it targeted an American phone number as a test for the FBI Contact Us Do you have more information about NSO Group, or other spyware companies? From a non-work device and network, you can contact Lorenzo Franceschi-Bicchierai securely on Signal at +1 917 257 1382, or via Telegram and Keybase @lorenzofb, or email. For years, NSO Group has claimed that its spyware cannot be used against American phone numbers, meaning any cell number that starts with the +1 country code. In 2022, The New York Times first reported that the company did “attack” a U.S. phone but it was part of a test for the FBI. NSO Group’s lawyer Joe Akrotirianakis confirmed this, saying the “single exception” to Pegasus not being able to target +1 numbers “was a specially configured version of Pegasus to be used in demonstration to potential U.S. government customers.” The FBI reportedly chose not to deploy Pegasus following its test. How NSO Group’s government customers use Pegasus NSO’s CEO Shohat explained that Pegasus’ user interface for its government customers does not provide an option to choose which hacking method or technique to use against the targets they are interested in, “because customers don’t care which vector they use, as long as they get the intelligence they need.” In other words, it’s the Pegasus system in the backend that picks out which hacking technology, known as an exploit, to use each time the spyware targets an individual. NSO Group’s headquarters shares the same building as Apple In a funny coincidence, NSO Group’s headquarters in Herzliya, a suburb of Tel Aviv in Israel, is in the same building as Apple, whose iPhone customers are also frequently targeted by NSO’s Pegasus spyware. Shohat said NSO occupies the top five floors and Apple occupies the remainder of the 14-floor building. “We share the same elevator when we go up,” Shohat said during testimony. The fact that NSO Group’s headquarters are openly advertised is somewhat interesting on its own. Other companies that develop spyware or zero-days like the Barcelona-based Variston, which shuttered in February, was located in a co-working space while claiming on its official website to be located somewhere else. NSO Group admitted that it kept targeting WhatsApp users after the lawsuit was filed Following the spyware attack, WhatsApp filed its lawsuit against NSO Group in November 2019. Despite the active legal challenge, the spyware maker kept targeting the chat app’s users, according to NSO Group’s research and development vice president Tamir Gazneli. Gazneli said that “Erised,” the codename for one of the versions of the WhatsApp zero-click vector, was in use from late-2019 up to May 2020. The other versions were called “Eden” and “Heaven,” and the three were collectively known as “Hummingbird.” NSO says it employs hundreds of people NSO Group’s CEO Yaron Shohat disclosed a small but notable detail: NSO Group and its parent company, Q Cyber, have a combined number of employees totalling between 350 and 380. Around 50 of these employees work for Q Cyber. NSO Group describes dire finances During the trial, Shohat answered questions about the company’s finances, some of which were disclosed in depositions ahead of the trial. These details were brought up in connection with how much in damages the spyware maker should pay to WhatsApp. According to Shohat and documents provided by NSO Group, the spyware maker lost $9 million in 2023 and $12 million in 2024. The company also revealed it had $8.8 million in its bank account as of 2023, and $5.1 million in the bank as of 2024. Nowadays, the company burns through around $10 million each month, mostly to cover the salaries of its employees. Also, it was revealed that Q Cyber had around $3.2 million in the bank both in 2023 and 2024. During the trial, NSO revealed its research and development unit — responsible for finding vulnerabilities in software and figuring out how to exploit them — made up the majority of a $52 million budget. Shohat also said that NSO Group’s customers pay “somewhere in the range” between $3 million and “ten times that” for access to its Pegasus spyware. Factoring in these numbers, the spyware maker was hoping to get away with paying little or no damages. “To be honest, I don’t think we’re able to pay anything. We are struggling to keep our head above water,” Shohat said during his testimony. “We’re committing to my [chief financial officer] just to prioritize expenses and to make sure that we have enough money to meet our commitments, and obviously on a weekly basis.” First published on May 10, 2025 and updated with additional details.

Amazon says it will work with Humain, the AI company recently launched by Saudi Arabia’s ruler, Mohammed bin Salman, to invest “$5 billion-plus” in a strategic partnership to build an “AI Zone” in Saudi Arabia. The AI Zone will include dedicated Amazon Web Services (AWS) AI infrastructure, servers, networks, and training and certification programs, according to a press release. Humain is pledging to develop AI solutions using AWS technologies, and to work with AWS on providing access to tools and programs for Saudi Arabia-based AI startups. AWS joins tech giants Nvidia, AMD, and others in partnering with Humain, which is funded by Saudi Arabia’s Public Investment Fund (PIF). American tech firms have looked to the PIF as a source of capital. Companies like Google and Salesforce have also recently worked with the PIF on AI-related projects and investments. President Donald Trump and several tech industry allies attended a U.S.-Saudi investment forum on Tuesday. Under a new Trump administration initiative, U.S.-based tech suppliers including the aforementioned Nvidia and AMD have been permitted to arrange deals with Saudi Arabian firms. Saudi Arabia has mandated AI companies and services in the kingdom store data locally, pushing vendors to put facilities there to avoid losing contracts. Both Google and Oracle have announced expansion plans in the region over the past year. Amazon early last March pledged to spend billions of dollars on data centers in Saudi Arabia. On Tuesday, the company said it is devoting around $5.3 billion to develop an AWS region in the kingdom scheduled to come online in 2026. The AI Zone commitments are an “additional investment” separate from the roughly $5.3 billion, Amazon said. It isn’t clear, however, if Amazon’s contribution to the AI Zone will draw on the tranche the company originally announced. TechCrunch has reached out to Amazon for clarification and will update this piece if we hear back. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW

Fabled startup investor and accelerator Y Combinator has some choice words for Google in an amicus brief it just submitted in the U.S.’s monopoly case against the search giant. In the brief, YC charged that Google is a “monopolist” that has “stunted” the U.S. startup ecosystem by making VC firms like itself hesitate to fund web search and AI startups in what it calls a “kill zone” around Google. “Google has chilled independent firms like YC from funding and accelerating innovative startups that could otherwise have challenged Google’s dominance,” YC wrote in the filing. “The result is a landscape that has been artificially stunted and stagnant.” YC’s brief says it’s currently seeking to fund startups developing question-based and agentic AI tools that could transform how people interact with information on the internet. But YC says there’s a “clear risk” that Google will use its monopoly power to slow down the future of those markets. “Google has effectively frozen the web search and text advertising markets for over a decade,” YC wrote. The brief, filed May 9, was spotted on X by VC Sheel Mohnot, the general partner of Better Tomorrow Ventures and a prolific social media poster. But YC isn’t calling for an immediate breakup of Google, as its CEO Garry Tan made clear in a reply to Mohnot. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW Rather, YC is arguing Google should curb practices it considers anti-competitive, like paying Apple billions of dollars to make Google the iPhone’s default search engine. It also wants Google to do things it argues would help startups, like opening up Google’s search index so others can train LLMs on it. For perspective, Google’s search algorithms have been its highly prized secret since its inception. For YC to ask the government to force Google to open it up to competitive LLMs is almost like demanding the government make Microsoft Windows open source, or forcing Amazon to freely deliver packages for competitors. If Google doesn’t implement such changes within a five year timeframe, then YC advocates for the government to force Google to divest or spin out parts of itself. YC CEO Tan characterized this idea in an X post as a “spinoff hammer” threat. He also posted that “we love Google” but wants “little tech” to succeed, too, in a separate X thread. YC released an amicus brief re: US v Google yesterday. We love Google and what it represents as a paragon of US-led tech and innovation. We also want to make sure the excesses of big tech make way for tomorrow's little tech. — Garry Tan (@garrytan) May 10, 2025 To recap, last year Google lost a massive antitrust case over its dominance of the search market. While Google appeals the decision, the U.S. government is mulling potential punishments (‘remedies’) that Google might be required to implement, such as spinning off Chrome. Those remedies are expected to be delivered by August 2025. YC’s stance may come as a surprise to those who have followed its latest partnerships with Google: most notably, Google Cloud gave YC startups access to a dedicated cluster of Nvidia GPUs last year. Google co-founder Larry Page also made a rare in-person appearance to speak at a YC event in December. Google has also acquired at least two YC-backed startups: Flutter in 2014, and Fridge in 2011. It also invested in YC startup Infisical through its Gradient fund in 2023. However, YC is also closely tied to OpenAI, which is now directly competing against Google on search. OpenAI’s CEO Sam Altman used to run YC, while OpenAI was the first group affiliated with YC Research. That’s something Mohnot pointed to on X, writing that the biggest beneficiary of YC’s proposed remedies, by far, would be OpenAI, rather than YC’s famously early stage startups, while commenting that the amicus brief “paints Google as more powerful than it is.” TechCrunch asked YC how it would respond to this critique, and whether it has any specific examples of areas that it probably would have funded had it not been for Google. So far, YC hasn’t responded to our comment request. Google didn’t respond to a request for comment about YC’s amicus brief, either. However, it argued in a blog post last year that the DOJ’s proposals are “radical and sweeping” and would hurt consumers, business, and developers.

As part of a broader app update, Airbnb on Tuesday introduced a new feature that allows travelers to book services and experiences, like getting a message, haircut, or chef-prepared meal, or taking part in some activity. These new offerings can be added to your stay, but they can also be booked independently, Airbnb says. For instance, you could book a tour in a city you’ve traveled to, but where you’ve chosen to stay elsewhere, the company suggests. This is not the first time Airbnb has tried such a feature — it paused an earlier version of experiences in 2023 to improve its core offering of stays. Now, the company once again wants customers to go beyond planning a trip. The move is meant to capitalize on the traffic Airbnb’s site already receives. During Airbnb’s Q1 2025 call, CEO Brian Chesky said that its service was accessed by over 1.5 billion devices in the last year, but pointed out that many people come to the site and don’t book a home. Image Credits: Airbnb It also brings Airbnb more directly in competition with other travel companies such as Tripadvisor, Booking.com, Viator, and GetYourGuide, and service providers like Yelp. Initially, Airbnb allows you to book services across 10 categories, including chefs, catering, prepared meals, photography, massages, spa treatment, personal training, hair, nails, and makeup. The services will be offered in 100 cities worldwide across eight countries. While users can take advantage of services where they’re staying, they will typically have to travel to a venue for the experiences. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW Initially, experiences will include cultural and museum tours; outdoor, watersport, and wildlife experiences; food tours and cooking classes; art workshops and shopping experiences; and workout, wellness, and beauty experiences. The company is launching experiences in 19 categories across 1,000 cities in the world. Image Credits: Airbnb Airbnb is also launching exclusive experiences on its platform called Airbnb Originals, which involve celebrity partnerships. These include things like pastry making at French Bastards bakery with chef Raphaelle Elbaz, or playing beach volleyball with Olympian Carol Solberg on Rio’s Leblon Beach. “These experiences and services are a way to experience a city like a local. We think that these are natural extensions to home [stays]. What’s great about Airbnb is that you get unique accommodations that you can’t find anywhere else. We think about our new offerings the same way,” Judson Coplan, VP of Product Marketing at Airbnb, told TechCrunch over a call. Coplan believes that Airbnb Experiences may inspire people to travel and take a trip they might not have thought about. Plus, he thinks that the new offerings can lead to people discovering new things to do in their own town. Airbnb will take a 15% cut from services and a 20% cut from experiences. However, users will just see one price when searching for or booking either category. The company says hosts in these categories will go through verification and quality checks that include their experience, online presence, education, and required licenses. Airbnb’s new app adds a social play The addition of services and experiences is part of a larger app update where both the apps for guests and hosts will be revamped to accommodate the new categories, among other changes. Guests will be able to explore and book from the Stay, Services, and Experiences categories. And, if travelers already have an upcoming trip booked, the app will suggest related experiences or services. On the other end, hosts will have a calendar and listings functions that help them manage bookings and offerings. Airbnb is also preparing to roll out social features later this year. The app today has a group messaging feature for trip planning, but it plans to add a group chat feature for people who have been to an experience together. That way, they can talk about the experience, share photos and videos, or make recommendations for another trip. In turn, Airbnb hopes this could lead people to want to travel more and use its platform to do so. The company says it’s developing privacy features around these social connections to create a safe experience for users. Image Credits: Airbnb The social features follow last year’s launch of an updated profile page, which allowed users to describe more about themselves, including where they live, languages they speak, and other facts about themselves. Coplan said that after that release, the number of completed profiles on the platform increased by 15x. “There was a huge increase in a number of people who wanted to share more about themselves. For us, that was a clue that travel is something about connections,” he explained. “Whether it’s your host, people you travel with, or people you meet along the way. There is especially, in this moment of time, a desire to connect with other people.” Scaling customer service with AI Mirroring comments made by Chesky, Coplan noted how Airbnb is using AI to provide customer service, and how, over time, its AI assistant will become a concierge that can handle trip planning and inspiration. Earlier this month, Chesky introduced Airbnb’s AI-powered customer service agent. Unlike the previous version of the bot, which redirected users to a support article, the assistant now provides an answer within the chat. The bot is initially available to English-speaking users in the U.S. and will be fully rolled out by month-end to this market. Later this year, the feature will become available to more countries and languages, Coplan told TechCrunch. Over time, he says Airbnb will add more features to its AI assistant, including personalized answers tailored to your trip and booking, and in-line action buttons for taking quick actions, such as canceling a booking. The updated Airbnb website and iOS and Android apps are rolling out to all users globally starting today.

At the Android Show, taking place ahead of Google I/O 2025, Google announced that it is adding new device-specific features to its Advanced Protection program, which is designed to protect public figures such as politicians and journalists from different digital threats, with the Android 16 release. The new features include a new way of storing device logs that are accessible only by the user, protection from spam calls, and an auto-restart feature for when the device is locked for a certain amount of time. These features are aimed at adding an extra layer of security for public figures and preventing their devices from being compromised. With the Android 16 release, the company is adding an intrusion logging feature for threat analysis. This feature stores your device logs with end-to-end encryption in the cloud. If there is suspicion around the device being compromised, these logs can be used to perform analysis. The company is also adding USB protection to allow only charging from a new USB connection while the device is locked. Google said devices under the Advanced Protection program will also automatically restart if the device is locked for 72 consecutive hours. TechCrunch reported last month that the feature was mentioned in the Google Play services update. What’s more, if you have Advanced Protection enabled, with Android 16, the system will prevent you from automatically reconnecting to Wi-Fi networks that are not secure. The program is also gaining a feature to privately process conversations on the device to try and detect a potential scam for an ongoing phone call. The company launched this feature for Pixel phones earlier this year. Check out how to watch the livestream and more from Google I/O. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW

TikTok is launching its first image-to-video AI feature, the company announced on Tuesday. The new feature is called “TikTok AI Alive” and allows users to turn static photos into videos within TikTok Stories. The feature is only accessible via TikTok’s Story Camera and uses AI to create short-form videos with “movement, atmospheric and creative effects,” TikTok says. For instance, if your static photo features a sky, clouds, and the ocean, TikTok could turn the photo into a video where the sky gradually shifts hues, the clouds start to drift, and you hear the sound of waves crashing. Or, you could animate a group selfie that highlights gestures and expressions. The launch of the new image-to-video features comes a few years after TikTok introduced an in-app text-to-image AI generator. While both Instagram and Snapchat also offer text-to-image AI features for creators, TikTok is now taking a step further by offering its users the ability to create videos from images. It’s worth noting that Snapchat has said it will soon allow creators to generate AI videos from images. Image Credits:TikTok AI Alive stories will have an AI-generated label to notify users that the content was created with AI. Plus, this content will have C2PA metadata embedded, which is a technical standard that helps others identify that the video is AI-generated, even if it’s downloaded and shared beyond TikTok. “We are always building with safety in mind, and the same goes for our AI innovations,” TikTok said in a blog post. “As this technology enables new forms of creative expression, it undergoes multiple trust and safety checks to protect our community. To help prevent people from creating content that violates our policies, moderation technology reviews the uploaded photo and written AI generation prompt as well as the AI Alive video before it’s shown to the creator.” TikTok notes that people can report videos that they think break the app’s rules, and that the app conducts a final safety check once a creator shares an AI Alive story. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW Creators can create an AI Alive video by opening the Story Camera and tapping the blue plus button on the top of the Inbox page or Profile page. From there, you can choose a photo from your Story Album. You will then see the AI Alive icon on the right side toolbar on the photo edit page.

WizardLM, a Beijing-based Microsoft AI research group, appears to have joined Tencent, the Chinese company that owns WeChat and blockbuster games like PUBG Mobile. In a post on X on Tuesday, Can Xu, a senior AI researcher who led a number of projects within WizardLM, said that he and the team left Microsoft to join Hunyuan, one of Tencent’s AI development orgs. Over the past few months, Hunyuan has released AI models including video and 3D object generators. WizardLM has already released a Hunyuan-branded model, in fact: Hunyuan-TurboS 0416. In an X post, Qingfeng Sun, who claims to be a “co-founder” of WizardLM, says that Hunyuan-TurboS 0416 outperforms “open” AI models like Google’s Gemma 3 series. It’s not clear how many researchers are still a part of WizardLM or when the team might’ve left Microsoft. We’ve reached out to both Tencent and Microsoft for comment. WizardLM, formed several years ago, has a bit of a strange history. Last April, the team, then under Microsoft, released a family of AI models, WizardLM-2, that it said was competitive with OpenAI’s GPT-4. But after only a day, Microsoft pulled WizardLM-2 from the web, saying that the models hadn’t undergone “toxicity testing.” “We accidentally missed an item required in the model release process — toxicity testing,” the WizardLM team wrote in a post on X on April 16, 2024. “We are currently completing this test quickly and then will re-release our model as soon as possible.” Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW The deletion came a little too late. Users quickly reuploaded the original WizardLM-2 models as well as customized, fine-tuned versions. Meanwhile, in a post on the AI dev platform Hugging Face, Hugging Face CEO Clément Delangue said that Microsoft’s move, which also wiped out other WizardLM models, harmed the Hugging Face community by breaking a number of open source projects. “[The WizardLM] models were collectively downloaded over a hundred thousand times a month,” Delangue wrote at the time. “We apologize for the inconvenience & are trying to get in touch with the author [and] Microsoft in order to try to find a good resolution for community members.” At Tencent, it seems WizardLM will keep doing what it was doing, more or less: developing and releasing AI models. Tencent recently restructured its Hunyuan team, establishing two new units and upping its AI infrastructure spending. Tencent attributed its 8% year-over-year growth in Q1 2025 to the company’s AI investments. Tencent has said that it plans to put 90 billion yuan, or around $12.49 billion, toward capital expenditures this year, much of which will fuel its AI efforts.

Federal railroad regulators are in talks with Elon Musk’s tunneling firm, The Boring Company (TBC), over a multi-billion Amtrak project, according to The New York Times. Citing sources familiar with the matter, the NYT reports that Federal Railroad Administration officials have spoken with TBC to learn whether the Musk-owned company could save the agency money on the Frederick Douglass Tunnel program. The project, which is being built to connect Baltimore to Washington and Virginia, was initially projected to cost $6 billion, but current estimates suggest it could cost as much as $8.5 billion. While TBC was one of several companies to be considered for a new engineering contract, the talks raise red flags about Musk’s many conflicts of interest as he splits his time between his companies, his role as President Trump’s right-hand man, and as overseer of DOGE, which has cut resources to agencies that regulate his businesses.

Android users will have more ways to find their devices and other items, Google announced on Tuesday during the Android Show, a week before Google I/O 2025. The company says its Find My Device feature, which allows Android users to locate lost phones and other devices, will become known as “Find Hub,” as it rolls out support for more partners, satellite-based finding capabilities, and airline partnerships. Initially introduced as an Android-based alternative to Apple’s Find My service in 2013, Find My Device expanded just over a year ago with the introduction of a crowdsourced finding network that could locate personal belongings via third-party Bluetooth trackers and tags, like those from Chipolo, Pebblebee, and others. Now, Google is unveiling more partners that will work with its service using built-in location tracking capabilities. Image Credits:Google This includes built-in luggage finding from brands like July and Mokobara, a way to locate lost skis with built-in Peak integration, and new Bluetooth tags aimed at families from Pixbee. Later this month, Android users will also be able to use ultra-wideband (UWB) technology to locate Motorola’s moto tags. The company says support for satellite-based finding will roll out later in 2025. Google also announced airline partnerships that will help travelers keep track of their luggage, following last year’s news that Apple’s Find My will work with over a dozen airlines. For Android users, new airline partners working with Google’s Find Hub include Aer Lingus, British Airways, Cathay Pacific, Iberia, and Singapore Airlines. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW

Microsoft is about to trim its worldwide workforce by 3%, CNBC reports. The company had around 228,000 employees worldwide as of June, it says, meaning more than 6,500 could be affected by the cuts. It would be one of the company’s largest staff reductions since it cut 10,000 people in 2023. A Microsoft spokesperson told TechCrunch, “We continue to implement organizational changes necessary to best position the company for success in a dynamic marketplace.” In April, Microsoft reported a solid quarter of $70.1 billion in revenue (up 13%) and net income profits of $25.8 billion (up 18%), which was even better than analysts expected. Even so, this new round of layoffs is expected to hit all levels, locations, and teams. The company also had layoffs back in January, though it said those were performance-based. This fresh round of layoffs is not performance-related, a spokesperson told CNBC. Thousands upon thousands of workers have been laid off from Big Tech companies in just the past year. Amazon and Meta also conducted layoffs in January.

After a week or so of rumors, the U.S. Department of Commerce (DOC) formally rescinded the Biden administration’s Artificial Intelligence Diffusion Rule days before it was set to go into effect. The Artificial Intelligence Diffusion Rule, which was introduced by former president Joe Biden in January and set to come into force on May 15, introduced limits on exporting U.S.-made AI chips to many countries for the first time while bolstering existing restrictions. On Tuesday, the DOC announced that it instructed staff not to enforce the Biden-era regulation. The DOC plans to issue a replacement rule in the future, likely focusing on direct negotiations with countries as opposed to blanket restrictions, according to reporting from Bloomberg. Biden’s proposed rule divided the world’s countries into three tiers, with each tier having its own level of restrictions. Tier 1 countries, like Japan and South Korea, would have continued to face no export restrictions; Tier 2 regions, which included countries like Mexico and Portugal, would have seen chip export limits for the first time; and Tier 3 countries, like China and Russia, would have had to contend with tightened controls. In lieu of new regulations, the DOC on Tuesday released some guidance for the industry. It reminded companies that using Huawei’s Ascend AI chips anywhere in the world violates U.S. export rules, warned about the potential consequences of letting U.S. AI chips be used to train AI models in China, and recommended ways to protect chip supply chains from diversion tactics. “The Trump Administration will pursue a bold, inclusive strategy to American AI technology with trusted foreign countries around the world, while keeping the technology out of the hands of our adversaries,” U.S. Secretary of Commerce for Industry and Security Jeffrey Kessler said in a statement. “At the same time, we reject the Biden Administration’s attempt to impose its own ill-conceived and counterproductive AI policies on the American people.”

During the Android Show leading up to Google I/O, Google on Tuesday offered a brief update on the adoption of the RCS (Rich Communication Services) protocol, an upgrade to SMS that offers high-resolution photos and videos, typing indicators, read receipts, improved group chat, and more. The company shared that the messaging standard now supports over a billion messages per day in the U.S. This metric is based on an average of the last 28 days, Google noted. The stat is notable because Google fought for years to get Apple to adopt support for RCS on iOS, allowing for better communication between Android and Apple devices. Previously, iPhone users who received texts from friends on Android had to deal with blurry videos and images, and couldn’t as easily manage group chats when their green-bubbled friends joined. Unlike with iMessage, group chats with Android users couldn’t be renamed, nor could people be added or removed, and you couldn’t exit when you wanted. Emoji reactions also didn’t work properly, leading to annoying texts to let you know how an Android user reacted, instead of just displaying their emoji reaction directly. That changed with the fall 2024 launch of iOS 18, when Apple finally added RCS support to its Messages app. Though the functionality has been upgraded, Apple still displays RCS chats as green bubbles, hoping to keep the stigma of being an Android user intact. This is particularly important among young people in the U.S., where demand for the blue bubbles has cemented the iPhone as teens’ most popular device. Surprisingly, Google did not share other stats around RCS, like those detailing the number of users more specifically or the carrier support. Google I/O 2025: What to expect, including updates to Gemini and Android 16

As serial entrepreneur Joel Milne founded, scaled, and then successfully sold mobile auto repair service startup RepairSmith to AutoNation, he was plagued by a persistent problem. The automotive retail industry has a communication problem. And it’s an expensive one. Thousands of dealerships and mechanic shops — each one with an array of software systems — lack a common language to make communication with manufacturers and other businesses easier. “We had this problem of, ‘how do you work with dealerships and shops and talk to them as you’re running around repairing cars and trying to refer them business or get parts from them?’” he said. “And it’s very fragmented, very difficult, very costly to build all these custom integrations with the different stores.” For instance, the average dealership relies on more than 40 different software systems, ranging from dealer management systems and customer relationship management tools to digital retailing, service, inventory, and payment processing platforms. Milne compared it to the financial services industry 20 years ago. Fintech company Plaid, which connects bank accounts to financial applications, helped close that communication gap. Milne wants to do the same, but for automotive retail. Milne is now founder and CEO of a new startup called AutoUnify that has built an API to allow dealerships and service shops to communicate in real-time with the manufacturers and software vendors that power their operations. AutoUnify has been operating quietly for nine months and is based in Santa Monica, California. After piloting with multiple customers in 2024, AutoUnify is now opening up its sales to the industry. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW AutoUnify is the latest startup to come out of a multi-year partnership between UP.Labs and Porsche. The startup has also raised $5 million in a round led by UP.Partners. Those funds will help the startup scale its workforce from the nine it employs today to about 20 by the end of the year. “Really the focus for the rest of the year is building the technology and building the sales pipeline,” he said. UP.Labs is not a venture firm, even though it emerged from, and operates in parallel with, UP.Partners. It’s not a corporate accelerator or incubator either. The company, which launched during UP.Summit 2022 in Bentonville, Arkansas, is structured as a venture lab with a new kind of financial investment vehicle. Up.Labs strikes partnerships with major corporations — Porsche was the first — and then works to identify the industry’s biggest problems and create startups with business models that will solve those pain points. In an unusual twist, these startups don’t simply serve the company, in this case, Porsche. To be successful, they have to be able to serve the broader market. Up.Labs has also locked in similar deals with Alaska Airlines and JB Hunt. Up.Labs CEO John Kuolt said they’ve uncovered some of the automotive industry’s biggest challenges while working with Porsche. To date, the companies have launched four startups, including Pull Systems, a software-as-as-service platform that provides performance management software to EV suppliers, manufacturers and operators, and Sensigo, which created an AI platform that allows service technicians to quickly diagnose problems in modern, software-defined vehicles. AutoUnify is its fourth startup, and one that stands out as one of the most critical and most difficult to solve, according to Kuolt. “It’s exactly the kind of breakthrough we build for: a company that not only tackles a technical challenge, but fundamentally reshapes how an entire industry operates,” he said.

Microsoft is about to trim its worldwide workforce by 3%, CNBC reports. The company had around 228,000 employees worldwide as of June, it says, meaning more than 6,500 could be affected by the cuts. It would be one of the company’s largest staff reductions since it cut 10,000 people in 2023. A Microsoft spokesperson told TechCrunch, “We continue to implement organizational changes necessary to best position the company for success in a dynamic marketplace.” In April, Microsoft reported a solid quarter of $70.1 billion in revenue (up 13%) and net income profits of $25.8 billion (up 18%), which was even better than analysts expected. Even so, this new round of layoffs is expected to hit all levels, locations, and teams. The company also had layoffs back in January, though it said those were performance-based. This fresh round of layoffs is not performance-related, a spokesperson told CNBC. Thousands upon thousands of workers have been laid off from Big Tech companies in just the past year. Amazon and Meta also conducted layoffs in January.

Google on Tuesday unveiled its new Android design language, Material 3 Expressive, at the Android Show ahead of Google I/O. The update is designed to make your phone more customizable and fluid, the tech giant says. Material 3 Expressive brings new springy animations that Google says will make your phone feel more fluid. For example, when you dismiss a notification, you will see a detach transition and feel a haptic rumble. You will see similar sorts of animation and feel haptics when you’re doing things like dismissing an app in your recent apps screen, fidgeting with the volume slider, or flinging down the shade. The update also subtly blurs the background to provide a sense of depth so you can stay in context when navigating your phone. “In 2021 we launched Material You as a big leap in design that focused around a user’s identity with experiences like dynamic color theming,” said Mindy Brooks, VP, Product and UX, Android Platform, in a briefing with reporters. “Now we’re building off the principles of Material You to bring our latest design update called Material 3 Expressive. Material 3 Expressive is an expansion of new components and capabilities designed to add emotion to product UIs, providing a more premium and engaging experience that is easier to use and creates a little more joy in key moments.” Image Credits:Google Android is also getting updated dynamic color themes and emphasized typography to allow users to customize their phone to their style and preferences. Plus, these visual customizations will be applied across Google apps, such as Google Photos and Gmail. In addition, you can now customize Quick Settings to add more of your favorite actions, such as Flashlight and Do Not Disturb. Android’s new Live Updates feature will also help you easily track progress notifications from select apps. For example, if you place an Uber Eats order, you will start to see a glanceable Live Update to see the real-time progress of your order. The feature is similar to iOS’s Live Activities feature. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW “We’re introducing Live Updates as a new feature that will highlight what you need to know so you can stay focused on what you care about,” Brooks said. “We’ve all been there. You’ve placed an order, and you’re trying to pay attention to conversations around the table. With glanceable live updates, you can easily track your order and get back to real life. These real time updates from your favorite delivery, rideshare, and navigation apps are coming as part of the redesign later this year.” All of these new design improvements will first come to Pixel devices later this year with Android 16. Image Credits:Google As for Wear OS watches, Material 3 Expressive’s design is more fluid with scrolling animations that trace the circular display. Just like on phones, the new design language will bring a sense of depth. Plus, Google is updating daily experiences like using the pin pad and controlling media with motion and responsive feedback. There will also be smoother transitions, along with dynamic color-theming to the watch so that the theme you choose for your watch face applies to the device. “We are excited to announce that we are bringing color theming to the watch,” Brooks said. “Now, the theme you choose for your watch face can theme the entire system in Google Apps so you can make it your own. And when it comes to getting things done, glanceability is one of the most important considerations for watch we’ve created a system of buttons that truly hug the display, making them that much more tappable and space efficient using an underlying polar grid.” Tiles are also going to be more engaging, making it easier to get quick access to information or actions, such as texting your favorite contacts or starting a workout. These changes will be rolling out to the Pixel Watch first later this year with Wear OS 6. Check out how to watch the livestream and more from Google I/O.

At the Android Show, ahead of Google I/O 2025, Google announced that it is adding new device-specific features to its Advanced Protection program, which is designed to protect public figures such as politicians or journalists from different digital threats, with the Android 16 release. The new features include a new way of storing device logs that are accessible only by the user, protection from spam calls, and an auto-restart feature for when the device is locked for a certain amount of time. These features are aimed at adding an extra layer of security for public figures and preventing their devices from being compromised. With the Android 16 release, the company is adding an intrusion logging feature for threat analysis. This feature stores your device logs with end-to-end encryption in the cloud. If there is suspicion around the device being compromised, these logs can be used to perform analysis. The company is also adding USB protection to allow only charging from a new USB connection while the device is locked. Google said devices under the Advanced Protection program will also automatically restart after 72 hours if the device is locked. TechCrunch reported last month that the feature was mentioned in the Google Play services update. What’s more, if you have advanced protection enabled, with Android 16, the system will prevent you from automatically reconnecting to Wi-Fi networks that are not secure. The program is also gaining a feature to privately process conversations on the device to try and detect a potential scam for an ongoing phone call. The company launched this feature for Pixel phones earlier this year. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW

Google is bringing Gemini, its generative AI, to all cars that support Android Auto in the next few months, the company announced at its Android Show ahead of the company’s 2025 I/O developer conference. The company says adding Gemini functionality to Android Auto and, later this year, to cars that run Google’s built-in operating system, will make driving “more productive — and fun” in a blog post. “This is really going to be, we think, one of the largest transformations in the in-vehicle experience that we’ve seen in a very, very long time,” Patrick Brady, the VP of Android for Cars, said during a virtual briefing with members of the media ahead of the conference. Gemini will surface in the Android Auto experience in two main ways. Gemini will act as a much more powerful smart voice assistant. Drivers (or passengers — Brady said they are not voice-matching to whoever owns the phone running the Android Auto experience) will be able to ask Gemini to send texts, play music, and basically do all the things Google Assistant was already able to do. The difference is users won’t have to be so robotic with their commands thanks to the natural language capabilities of Gemini. Gemini can also “remember” things like whether a contact prefers receiving text messages in a particular language, and handle that translation for the user. And Google claims Gemini will be capable of doing one of the most commonly-paraded in-car tech demos: finding good restaurants along a planned route. Of course, Brady said Gemini will be able to mine Google listings and reviews to respond to more specific requests (like “taco places with vegan options”). The other main way Gemini will surface is with what Google is calling “Gemini Live,” which is an option where the digital AI is essentially always listening and ready to engage in full conversations about … whatever. Brady said those conversations could be about everything from travel ideas for spring break, to brainstorming recipes a 10-year-old would like, to “Roman history.” Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW If that all sounds a bit distracting, Brady said Google believes it won’t be. He claimed the natural language capabilities will make it easier to ask Android Auto to do specific tasks with less fuss, and therefore Gemini will “reduce cognitive load.” It’s a bold claim to make at a time when people are clamoring for car companies to move away from touchscreens and bring back physical knobs and buttons — a request many of those companies are starting to oblige. There’s a lot still being sorted out. For now, Gemini will leverage Google’s cloud processing to operate in both Android Auto and on cars with Google Built-In. But Brady said Google is working with automakers “to build in more compute so that [Gemnini] can run at the edge,” which would help not only with performance but with reliability — a challenging factor in a moving vehicle that may be latching onto new cell towers every few minutes. Modern cars also generate a lot of data from onboard sensors, and on some models, even interior and exterior cameras. Brady said Google has “nothing to announce” about whether Gemini could leverage that multi-modal data, and that “we’ve been talking about that a lot.” “We definitely think as cars have more and more cameras, there’s some really, really interesting use cases in the future here,” he said. Gemini on Android Auto and Google Built-In will be coming to all countries that already have access to the company’s generative AI model, and will support more than 40 languages.

During the Android Show on Tuesday, leading up to Google I/O, Google announced that it’s bringing Gemini to Wear OS smartwatches and Google TV. The announcement is part of Google’s plans to replace Google Assistant with Gemini across all of its devices and platforms. Tuesday’s announcement doesn’t come as a surprise, as users have been expecting Google to bring its AI assistant to Wear OS smartwatches and Google TV. The tech giant says this expansion will make its AI assistant accessible beyond phones to allow people to use it when they’re doing things like going on a run or relaxing on the couch. Wear OS smartwatch users will get access to Gemini on their wrist, which Google says will be useful in situations where you can’t use your phone, like when you’re covered in flour while cooking or in the middle of a bike ride. Google notes that you can talk to Gemini naturally, which means that you don’t have to get the words exactly right. Image Credits:Google For example, if you’re in a locker room at your gym and you already put your phone away, you can use Gemini on your smartwatch to set reminders, like “Remember, I’m using locker 43 today.” Plus, since Gemini can connect to your apps, you can ask about the restaurant that your friend just emailed you about, without having to pause your workout or use your phone. Gemini is coming to Wear OS smartwatches in coming months. Google says Gemini on Google TV will allow users to do things like get personalized recommendations for content, like asking for a list of action movies that are age-appropriate for kids, for example. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW Image Credits:Google Gemini will also be able to answer questions. For example, your kids can ask the assistant about the solar system, after which Gemini could pull up the perfect educational YouTube video to help them learn more about their various interests. Gemini will launch on Google TV later this year. The company is also integrating Gemini to Android XR, its new platform for Extended Reality (XR) devices like headsets and glasses. Google announced back in December that Android XR will first launch with the Samsung-built Project Moohan headset, which will be available for purchase sometime this year. Image Credits:Google It’s not surprising that Google is bringing Gemini to Android XR, as the company already previously said that users will be able to control the headset with Gemini and ask questions about the app and content that they are looking at. “When the first headset launches later this year, Gemini can help you do things like planning a vacation by surrounding you with videos, maps and local tips, and this helps you create an entire itinerary in minutes within this very rich, immersive experience,” said Guemmy Kim, senior director of product and user experience on the Android team, in a briefing with reporters. As part of Tuesday’s announcement, Google shared that Gemini is coming to Android Auto in the coming months. This expansion will allow users to do things like ask Gemini to find a charging station that’s, for example, on the way to the post office that’s also near a park. Plus, you can ask it to summarize the messages you receive or give you a rundown of the news.

At the Android Show on Tuesday, ahead of Google I/O, Google announced new security and privacy features for Android. These new features include new protections for calls, screen sharing, messages, device access, and system-level permissions. With these features, Google aims to protect users from falling for a scam, keep their details secure in case a device is stolen or taken over by an attacker, and enhance device-level security for various attacks. Phone scammers often ask users to take actions like tapping on unsafe links or downloading unknown apps. In order to protect users, Google is blocking some actions and warning users of a potential scam while they are on a call with someone not in their contact list. Image Credits:Google For Android 16, these actions include side-loading an app for the first time from a web browser, messaging app, or other sources that have not been verified by Google, and granting accessibility permission to an app so that a scammer can take control of the device. The company is also preventing users running Android 6 or later from disabling Google Play Protect, which scans the device for harmful apps while they are on a call. Google is adding screen-sharing protection as well by reminding users to stop sharing the screen after a call ends. The company is also testing a new warning screen with select banks in the U.K. to prevent fraud through screen-sharing. When users on devices running Android 11 or later open a partner bank’s app while sharing the screen while on a call with an unknown number, the device will show them a new warning screen about a possible scam with a button to quickly end the screen sharing. Image Credits:Google Messages protection Google is enhancing protection against scams in Google Messages after launching the feature in March. The feature uses on-device AI to alert users of a potential scam based on the conversation. The company is now bolstering user security by having the tool to detect more types of fraud, including crypto, gift card, toll road, and other billing fees, financial impersonation, and technical support. Image Credits:Google What’s more, the company is adding verification keys to the Google Contacts app, which will help users authenticate that the person on the other end is the intended receiver. Users can verify their contacts by having them scan a QR code or match numbers displayed on the screen. Techcrunch event Join us at TechCrunch Sessions: AI Secure your spot for our leading AI industry event with speakers from OpenAI, Anthropic, and Cohere. For a limited time, tickets are just $292 for an entire day of expert talks, workshops, and potent networking. Exhibit at TechCrunch Sessions: AI Secure your spot at TC Sessions: AI and show 1,200+ decision-makers what you’ve built — without the big spend. Available through May 9 or while tables last. Berkeley, CA | June 5 REGISTER NOW Verified keys mean that your conversation with your contact on Google Messages is end-to-end encrypted and secured. Google said that if an attacker starts controlling a phone number through a SIM swap attack and messages you through a new device, the Google Contacts app will show the verification status as unverified. This feature will be available for Google Messages later this summer for users running Android 10 or later versions. Image Credits:Google Theft protection measures Earlier this year, Google rolled out Identity Check protection to Pixel and Samsung devices with OneUI 7. The feature requires users to use biometric authentication to change critical settings like changing your device PIN or biometrics, disabling theft protection, or accessing Passkeys when the user is not at one of the trusted locations they have added. The company is now making this feature available to other device makers with Android 16. Later this year, Google will add better protection for Factory Reset, restricting all kinds of functions on the device that are reset without authorization of the previous lock pattern/PIN or Google account credentials. This essentially makes a stolen device hard to use. Image Credits:Google To prevent someone from locking your device remotely, the company is adding a security challenge question to prevent unauthorized access. With Android 16, the company will also hide one-time passwords if the device is not connected to Wi-Fi and hasn’t been unlocked recently. Image Credits:Google Additional protections The company is adding new features to its Google Play Protect live detection program as well, by detecting unsafe apps that have hidden or changed icons. This feature will be available in the coming months for users running Android 6 and later versions. The company said it is now applying a new set of on-device rules to catch more categories of malicious apps. Google is adding new measures to bolster its Advanced Protection Mode, to protect public figures with new on-device features. The company is also debuting a new Find My Hub to keep track of items, friends, and family.

Apple Music is introducing a new “audio wellness” collection of songs called Sound Therapy. Designed in partnership with Universal Music Group (UMG), the feature combines songs that subscribers already know with other sound waves that promote focusing, relaxing, or sleeping. Songs may include auditory beats or colored noise, sound waves where the intensity varies at different frequencies that are categorized by colors. For instance, pink noise — a variation akin to rain or wind — can be combined with delta waves to promote sleep. Image Credits:Apple The Focus category in Sound Therapy, meanwhile, leverages gamma auditory beats combined with white noise. And songs in the Relax group use theta auditory beats to promote winding down and letting go. The collection features songs that listeners already know, like those from artists such as Imagine Dragons, Katy Perry, Kacey Musgraves, Ludovico Einaudi, AURORA, Jhené Aiko, Chelsea Cutler, and Jeremy Zucker. These tracks were turned into extended, instrumental versions by producers and audio engineers at UMG’s music wellness venture called Sollos, based in London. The addition joins the recently launched Apple Music Chill radio station, offering sounds that are also interspersed with mindful moments and calming stories.