WireGuard, the major software project and VPN that underpins popular security software including Mullvad and others, has found itself locked out of a key part of its Microsoft developer’s account and unable to ship software updates to Windows users.

Jason Donenfeld, the creator of the open source WireGuard VPN software, told TechCrunch that he has been locked out of his Microsoft developer account, and as a result cannot sign drivers or ship updates for WireGuard for Windows users, which are critical for its software to run. Donenfeld said in a post on X on Wednesday that the account termination stopped a WireGuard update from shipping.

It’s the second such incident of a high-profile and widely used open source project being shut out from its customers due to a seemingly abrupt account termination from Microsoft, with popular encryption software VeraCrypt facing a similar circumstance. Both developers said Microsoft locked them out of their accounts without first alerting them.

In the case of VeraCrypt, which is used by hundreds of thousands of users to encrypt files and operating systems, its developer Mounir Idrassi told TechCrunch that being locked out of his account means he is unable to update the software in time for a crucial certificate authority expiry, which he said may prevent some users from booting up.

Donenfeld, the WireGuard developer, told TechCrunch in an email: “If there were a critical vulnerability to fix right now — there isn’t! I just mean hypothetically — then users would be totally exposed.”

WireGuard is an open source VPN software used around the world to connect devices over the internet. WireGuard’s code is highly popular for its simplicity and security, as it serves as the foundation of many VPN implementations and commercial services that rely on its code, like Proton and Tailscale.

Donenfeld told TechCrunch in an email that he has spent the past few weeks modernizing WireGuard’s Windows code and was ready to send a copy update to Microsoft for checks before it can ship out to users, but was met with an “access restricted” error when logging into the developer portion of his Microsoft account.

Despite going through the process to verify his driver’s license or passport with Microsoft (the third party Microsoft uses for verification said he was “verified”), Donenfeld said his access was still suspended.

Donenfeld told TechCrunch that he found a page on Microsoft’s website saying that the company had been carrying out “mandatory account verification for all partners in the Windows Hardware Program who have not completed account verification since April 2024,” but that the verification program had since closed.

Microsoft’s Windows Hardware Program allows developers like Donenfeld and VeraCrypt’s Idrassi to “deploy hardware and device drivers for Windows PCs and other devices.” The ability to develop and release drivers for Windows users is restricted to known and vetted developers, as drivers can grant vast access to an operating system and its data and are known to be abused by hackers for that reason.

That account verification process meant that developers were required to upload their government-issued ID before they were allowed to publish potentially highly sensitive code to the broader Windows user base.

“Microsoft never sent me any notification at all about this. I’ve looked in every inbox in every spam folder in every mail log, and zero, nothing, zilch,” Donenfeld said.

The Windows Hardware Program’s verification program has “now concluded” and developers who have not uploaded their documents had their accounts “suspended,” the page reads, meaning that these accounts can no longer send updates.

Donenfeld said that he was referred to Microsoft’s executive support team, which handles customer service and account requests for high-profile individuals, which confirmed his appeal had been received but that they had to wait as long as 60 days for review.

By late Wednesday, there was a glimmer of hope in Donenfeld’s case. He told TechCrunch that he was finally in contact with Microsoft and that hopefully the issue would be resolved soon.

Microsoft did not immediately comment when reached by TechCrunch.

Donenfeld and Idrassi are not alone, with the account lockout issues affecting others as well.

Windscribe, a maker of VPN and other consumer privacy tools, said in a post on X that it had also been locked out of its Partner Center account. The company said it had a verified account for over eight years in order to sign its drivers.

“We’ve been trying to resolve this for over a month, and getting nowhere. Support is non-existent,” Windscribe said in its post. “Anyone know a human with a brain that still works at Microsoft and can help?”

The death of a duck in the Austin, Texas enclave of Mueller Lake has neighbors raising concerns about autonomous vehicles and whether they belong there.

While humans are responsible for killing animals with their cars all the time, this incident has brought negative attention to the new technology. Local media picked up on the duck incident after a resident posted in a Mueller neighborhood Facebook group that an Avride autonomous vehicle (with a human safety operator behind the wheel) ran over and killed a duck, and did not stop afterwards. “It didn’t slow down or hesitate at all, just steamrolled through,” the post, which KXAN reported on, reads.

Residents’ familiarity with this particular duck, which was nesting in a pot located outside of a local Italian eatery, has added to the outrage and mistrust of the autonomous vehicle technology. For those concerned about the future of the duck’s eggs, local residents have them in an incubator, Axios’ Austin reports.

An Avride spokesperson confirmed with TechCrunch that the vehicle was in autonomous mode at the time. Avride hasn’t paused testing on public roads altogether. However, the company has adjusted its area of operations by excluding certain streets around the lake in the neighborhood where the incident with the duck occurred, according to spokesperson Yulia Shveyko.

The resident also claimed in their post that the vehicle failed to stop at a stop sign. Avride told TechCrunch it did not find evidence to support that claim. The vehicle came to complete and appropriate stops at all relevant stop signs.

Shveyko said the team has reviewed vehicle data and behavior, including replaying the scene multiple times in simulation. Avride is now evaluating potential improvements to the technology to help avoid similar situations in the future, she said. Notably, this includes running a series of controlled experiments in simulation to ensure that any changes do not negatively impact the vehicle’s safety performance in other scenarios.

Avride isn’t the only company testing or commercially deploying autonomous vehicles in the city. Zoox has been testing in the city. Tesla and Waymo, in partnership with Uber, also operate a commercial robotaxi service in parts of Austin.

Is Poke an OpenClaw for the rest of us? That’s the idea coming from a new startup offering an AI agent that you can access via iMessage, SMS, Telegram, and, in some markets, WhatsApp.

The AI agent Poke launched publicly in March, allowing consumers to access a personal assistant that can take action on their behalf through a familiar interface. Today, Poke can help with everyday needs, like daily planning, managing your calendar, tracking your health and fitness, controlling your smart home, editing your photos, and more, all via text message.

Image Credits:Poke/The Interaction Company of California

While you may still interact with a general-purpose AI chatbot like ChatGPT or Claude when you have questions or want to do research, you’d turn to Poke when you want to get something done quickly, or when you want to automate some task to save you time.

For instance, you could ask Poke to alert you to specific emails (like those from your family or your boss), or remind you in the morning if you need to take an umbrella with you. It could help you track your health and fitness goals, or let you know the score to last night’s game. Poke could send daily medication reminders, or catch you up on the day’s news, and more, since users can write their own automations in plain text and then share them with friends.

Backed by Spark Capital, General Catalyst, and other angels, the 10-person startup has more recently added another $10 million to its coffers, on top of last year’s $15 million seed round. It’s now valued at $300 million, post-money.

Starting today, personal superintelligence is just one tap away.



No download, no signup.



Text Poke for free now:https://t.co/VIWYU64dUI 🌴



—

0:00 – What's Poke?

0:50 – Introducing Poke Recipes

1:25 – Create a Recipe in 10 seconds

1:43 – Earn on Poke

2:44 – Build with npx… pic.twitter.com/LHLFRVgahk — Poke (@interaction) March 19, 2026

The tool arrives as demand for agentic AI systems is spiking, leading OpenAI to snap up OpenClaw’s creator, and Nvidia’s CEO Jensen Huang to warn that every company needs its own OpenClaw strategy when announcing Nvidia’s enterprise-grade alternative.

But for those less technically inclined, the prospect of having to install software through the terminal, manage dependencies, and troubleshoot errors is daunting. Plus, systems like OpenClaw raise security concerns due to its deep system access.

For many people, then, OpenClaw and other agentic systems still feel out of reach. The team behind Poke wants to change that.

Image Credits:Poke/The Interaction Company of California

Marvin von Hagen, co-founder of The Interaction Company of California, the Palo Alto-based startup behind the new AI agent, tells TechCrunch that Poke emerged from watching how beta testers were using the company’s earlier product, an AI assistant for email, built around a year ago.

“What we noticed there was that people wanted to use Poke for everything… Even though it was only meant for email, people started asking Poke to remind them to take their medication. They asked Poke about sports results — ‘Hey Poke, tell me every morning if I need a jacket or not,’” explains von Hagen. “And at that time, we didn’t have a lot of this functionality, but we noticed how we needed to become general-purpose much more quickly, because people just like the personality and the humanness of it so much.”

The team then partially pivoted and focused on making Poke more useful, proactive, and more personable.

Unlike OpenClaw, getting started with Poke is easy. You simply visit Poke.com, click “Get Started,” and enter your phone number. There’s no app to install as the assistant operates over text messaging.

Image Credits:Poke/The Interaction Company of California

Under the hood, Poke turns to the AI model that best fits the task, whether that’s a model from one of the big AI providers or an open source model.

“I think this is also one of our main strengths in the long run: that almost all of our competitors are just big tech and labs that are bound to a specific provider. Like Meta AI will only ever be able to use Meta models, and ChatGPT will only ever be able to use OpenAI models,” von Hagen points out.

To work over messaging platforms like iMessage, Poke also leverages Linq, a solution that enables AI assistants to live within messaging apps. The app can run through SMS and Telegram, too, but WhatsApp support is currently limited as Meta barred other general-purpose chatbots last fall.

That could change, however. Regulators from the EU, Italy, and Brazil opened antitrust probes to fight this decision, which has brought Poke back to Brazil. It will hopefully also allow Poke to work on WhatsApp in the EU when Meta brings the costs down. (Meta has seen pushback over the high fees it’s charging — von Hagen says it’s a form of “malicious compliance” that he believes will soon be addressed.)

Image Credits:Poke/The Interaction Company of California

At launch, Poke offers a variety of “recipes” — or pre-made tools that help you automate various aspects of your life or work. These span categories like health and wellness, productivity, finance, scheduling, travel, home, school, email, community, and, for those who are technical, developer tools. Installing them requires a click of a button and then a standard authorization process, if needed.

These recipes are designed to work with apps and services you already know, like Gmail, Google Calendar, Outlook, Notion, Linear, Granola, and others. There are health and fitness “recipes” that work with Strava, Withings, Oura, Fitbit, and more, as well as those that work with smart home devices from companies like Philips Hue and Sonos.

Developers using Poke can also automate parts of their workflow via integrations with tools like PostHog, Webflow, Supabase, Vercel, Devin, Sentry, GitHub, Cursor Cloud Agents, and others.

Poke’s security model is multi-layered and includes regular penetration testing, security checks, various tools, and limiting permissions for both agents and human employees. By default, the team can’t see anything inside the tokens, unless the user manually opts to provide access to a log file or analytics by flipping a switch in their settings to opt to share this information. (TechCrunch has not performed its own security audit, to be clear.)

Image Credits:Poke screenshot/TechCrunch

Over the past couple of weeks, Poke’s users have created thousands more recipes and automations, which the company plans to add to its recipes directory for discovery in the near future. It’s also encouraging creators to build these shareable recipes by offering to pay somewhere between 10 cents and a dollar (based on geography) for every user who signs up for Poke via the recipe.

Image Credits:Poke.com screenshot/TechCrunch

The cost to use Poke is surprisingly affordable: it’s free to start, then pricing is flexible. During the beta tests, users actually had to negotiate with the AI agent what price they’d pay per month, which ranged between $10-$30 — or so Poke told us in response to this question.

Von Hagen says that, now, pricing is based on how the AI agent is being used. If you’re asking for things that don’t require real-time data, you could probably use Poke for free. What costs Poke money is real-time inference, like automations that run on every incoming email or real-time flight check-ins. To set prices, the company gave Poke guidance on how expensive things are, which allows it to determine personalized pricing.

While the company has managed to make Poke more efficient to reduce costs, the goal right now isn’t profitability, von Hagen notes.

Image Credits:Poke screenshot/TechCrunch

“We really don’t want to make money, but we really want to grow. We want to build a product for a billion people and monetization is really secondary,” he says. “The goal for the next weeks and months now is to bring Poke into everyday life.” To do so, it will look to creators and influencers to showcase how they’re using Poke.

I built "tastebuds", an MCP server on @interaction’s Poke Recipes to crowdsource opinions on food.



When you’re looking for a place to eat, Poke recommends food spots in your area that other Poke users have enjoyed before by storing their reviews, all running quietly in the… — Dani (@daniticow) April 8, 2026

The company, co-founded by Felix Schlegel, isn’t sharing how many customers have signed up, beyond noting that the figure has 10x’ed over the past couple of months. (However, we did spot Poke at the top of Vercel’s AI Gateway leaderboard, for whatever that’s worth.)

In addition to its main institutional investors, Spark Capital and General Catalyst, the startup has attracted the attention of a number of angels, including John and Patrick Collison (Stripe founders), Jake and Logan Paul, Logan Kilpatrick from DeepMind, Joanne Jang of OpenAI, and Scott Wu and Walden Yan (Cognition founders).

It also included Vercel co-founder Guillermo Rauch, PayPal co-founder Ken Howery, Dropbox co-founder Arash Ferdowsi, Mercor co-founder Brendan Foody, Hugging Face co-founder Thomas Wolf, Flapping Airplanes co-founder Ben Spector, and several others.

Canva announced on Wednesday the dual acquisition of Simtheory, an AI collaboration and agent management platform, and Ortto, a customer data and marketing automation company. The financial terms of the deals were not disclosed. Canva says the acquisitions are part of its continued investment in AI and marketing infrastructure.

Both companies were founded by Chris and Mike Sharkey, previously founders of vacation rental service Stayz, which was acquired by Fairfax Media. The pair will join Canva in leadership roles across the company’s AI and marketing technology teams.

Canva says the acquisitions add strengths in agentic AI, data infrastructure, marketing automation, and customer engagement, helping Canva evolve from a design tool into a system where teams do all of their work from start to finish. The additions extend Canva’s capabilities across the full workflow, from early ideas to campaign scaling and measurement.

Image Credits:Simtheory

Simtheory’s platform helps teams use AI to build assistants that understand their business, work across tools, and handle real tasks. Simtheory lets teams apply the latest models across diverse use cases and set up agentic workflows tailored to their needs.

On the other hand, Ortto combines a customer data platform with marketing automation, letting teams design and run journeys across email, SMS, push, in-app messaging, forms, and surveys in one system. Its event-driven architecture and no-code integrations make it easy to connect and activate data in real time, Canva says. The platform is used by more than 11,000 customers in 190 countries.

“Simtheory accelerates our evolution from a design platform with AI tools to an AI platform with design and productivity tools at its core,” said Canva co-founder and COO Cliff Obrecht, in a press release. “At the same time, Ortto strengthens our ability to power the entire marketing and content lifecycle through Canva Grow, from planning and creating to publishing and optimising across every channel.”

Image Credits:Ortto

Canva Grow is the company’s tool for asset creation and performance measurement.

Techcrunch event This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. San Francisco, CA | REGISTER NOW

These latest acquisitions come as Canva has been on a bit of an acquisition spree. Two weeks ago, the company acquired digital outdoor advertising startup Doohly. Six weeks ago, Canva announced the dual acquisition of startups Cavalry, which works on animation, and MangoAI, which works on improving ad performance. Before that, Canva acquired marketing intelligence startup MagicBrief in January 2025.

Canva closed 2025 at $4 billion in annualized revenue with more than 265 million users and 31 million paid users. The company also saw a 20% increase in monthly active users.

AWS CEO Matt Garman said Amazon’s recent $50 billion investment in OpenAI, after its long partnership including $8 billion of investment in Anthropic, is the type of conflict of interest the cloud giant is used to handling.

Garman has worked at Amazon since he was a business school intern in 2005, before the launch of AWS in 2006, he told the audience of the HumanX conference taking place this week in San Francisco.

When asked about the inherent conflict of working closely with two AI model companies that are fierce (and, arguably, sometimes petty) competitors, he said it’s not a problem. Because AWS itself often competes with its partners, it has a lot of direct experience with such competition, he explained.

In AWS’s earliest years, it knew it couldn’t build every cloud offering itself, so the unit partnered with others.

“We also knew that we would have to compete with our partners, because technology is interconnected,” Garman recounted. “So, for a very long time, we’ve built this muscle up of how we go to market with our partners,” he continued. “But we also may even have first-party products that compete with them, and that’s okay, and we’ve promised them we won’t give ourselves unfair competitive advantage.”

Today, the world is used to Amazon competing with those who sell on its cloud. Even one of AWS’s biggest rivals, Oracle, sells its database and other services on AWS. But it was a radical idea back in 2006, when technology partners took pains never to compete with the partners that helped them succeed.

Still, Amazon is hardly a trailblazer in discarding investor loyalty and conflict-of-interest commitments in the wild, money-grabbing world of AI. When Anthropic announced its latest $30 billion round in February, it included at least a dozen investors who were also backing OpenAI. This included OpenAI’s main cloud partner, Microsoft.

Techcrunch event This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. San Francisco, CA | REGISTER NOW

For AWS, making a huge investment in OpenAI to gain its model for its customers (and as a technology development partner) was almost a matter of life and death. Both models were already available on Microsoft’s cloud, AWS’s biggest rival.

The cloud giants are also working to keep themselves front and center by offering AI model-routing services. Those services allow their customers to automatically use different models for various tasks as a way to maximize performance and reduce costs. As Garman explained, one model might be ideal for planning, another for reasoning, and a cheaper model for easier tasks, like code completion. “I think that is where the world will go,” Garman said.

That is also how Amazon, and Microsoft for that matter, will slip their own homegrown models into usage — that old competing-with-your-partners situation, again.

All’s fair in love and AI these days.

Amazon is preparing to end support for Kindle devices released before 2012 — cutting them off from the Kindle Store and making them virtually unable to load any new content. The company set a cutoff date for May 20, 2026.

Amazon sent an email to customers this week announcing it would end support for the devices, according to screenshots of the message posted to social media and Reddit from customers, and confirmed by Amazon. According to the message, Kindle e-readers and tablets released in 2012 or earlier will lose access to Kindle Store functions after the deadline. That means affected devices will no longer be able to buy, borrow, or download new books directly.

After May 20, users will only be able to use their devices to read content that’s already downloaded. Once an older device is deregistered or reset to factory settings, it cannot be re-registered after May 20.

Amazon’s list of impacted models includes the original Kindle (1st and 2nd generation), Kindle DX and DX Graphite, Kindle Keyboard, Kindle 4 and 5, Kindle Touch, and the first-generation Kindle Paperwhite.

“These models have been supported for at least 14 years — some as long as 18 years — but technology has come a long way in that time, and these devices will no longer be supported moving forward,” Jesse Carr, an Amazon spokesperson, said in a statement to TechCrunch. “We are notifying those still actively using them and offering promotions to help with the transition to newer devices.”

Users can continue to access their libraries through newer devices, the Kindle mobile app, or Kindle for Web, the Amazon spokesperson said.

Several longtime users voiced frustrations about the move to end support for older Kindles, many of which are in excellent condition and have not seen any operational degradation. On Reddit, multiple users said they still rely on these older devices and see no reason to replace them. For some, it felt like another example of tech companies discontinuing support for perfectly functional products, pushing customers toward upgrades while contributing to growing electronic waste. One example of this is Google, which ended support for its older Nest thermostats in 2025. According to the United Nations Institute for Training and Research (UNITAR), global e-waste is estimated to reach 82 million tonnes by 2030, a 32% jump from 2022.

Techcrunch event This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. San Francisco, CA | REGISTER NOW

One Reddit user wrote, “I’ve had my Kindle for years, but it still works perfectly and continues to serve me well. How wasteful is it to make a product practically unusable in order to force people to buy a newer model.”

For some, the change meant it might be time to get an alternative e-ink reader, like the Boox Palma or Vivlio e-reader.

The decision to essentially kill older Kindle devices isn’t entirely unexpected, as Amazon has gradually reduced support for its legacy devices. In 2016, the company required Kindle e-reader models from 2012 and earlier to install updates just to maintain access to the Kindle Store, an early sign that long-term support for these devices would eventually come to an end.

Meta released an AI model on Wednesday called Muse Spark, which marks its “first step” toward an “overhaul of [its] AI efforts.”

Muse Spark is the inaugural model to come out of Meta Superintelligence Labs, which was created last year because CEO Mark Zuckerberg was reportedly unhappy with the progress of Meta and its Llama models and how they lagged behind OpenAI’s ChatGPT and Anthropic’s Claude. Meta recruited former Scale AI co-founder and CEO Alexandr Wang to lead Meta Superintelligence Labs and invested $14.3 billion in the data labeling company for a 49% stake.

Now, it’s time for Zuckerberg to see if his reconfigured AI team can woo users.

Muse Spark, which is now available on the web and the Meta AI app, is expected to improve over time. The company plans to roll out a “Contemplating” mode, which allows it to tackle more complex problems. Meta’s model uses multiple AI agents at once to work on the same problem, which it says will generate faster results for its Contemplating mode.

“To spend more test-time reasoning without drastically increasing latency, we can scale the number of parallel agents that collaborate to solve hard problems,” the company wrote.

Meta’s competitors have historically placed these more capable models behind a paywall. It’s unclear if Meta will follow the same strategy.

The company has jumped on one AI industry trend, though. Meta said in its blog post that Muse Spark could be applied to help users with health questions, something that competitors are also working on.

Techcrunch event This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. San Francisco, CA | REGISTER NOW

Meta’s push into health — and even the more basic function of logging into Muse Spark — could raise privacy concerns. Muse Spark users will need to log in with an existing Meta account such as Facebook or Instagram in order to use it. Meta doesn’t explicitly say that personal information from a Facebook or Instagram account will be used by the AI. But it is likely, considering that Meta generally trains on public user data and the company has positioned Muse Spark as a personal superintelligence product.

Meta also said that Muse Spark performs especially well with visual STEM questions which can lead to “interactive experiences like creating fun minigames or troubleshooting your home appliances.”

Aside from its investment in Scale AI and hiring of Wang, Zuckerberg’s company has recruited researchers from OpenAI, Anthropic, and Google. The upshot: If Meta’s going to be a real competitor in the AI industry, it’s now or never.

“Looking ahead, we plan to release increasingly advanced models that push the frontier of intelligence and capabilities, including new open source models,” Zuckerberg wrote on Threads. “We are building products that don’t just answer your questions but act as agents that do things for you.”

Security researchers say they have identified a hack-for-hire group targeting journalists, activists, and government officials across the Middle East and North Africa. The hackers used phishing attacks to access targets’ iCloud backups and messaging accounts on Signal, and deployed Android spyware capable of taking over the targets’ devices.

This hacking campaign highlights a growing trend of government agencies outsourcing their hacking operations to private hack-for-hire companies. Some governments already rely on commercial companies that develop spyware and exploits used by police and intelligence agencies to access data on people’s phones.

Researchers from the digital rights organization Access Now documented three instances of attacks over 2023 through 2025 against two Egyptian journalists, and a journalist in Lebanon whose case was also documented by digital rights organization SMEX.

Mobile cybersecurity company Lookout also investigated these attacks. The three organizations collaborated with each other and published separate reports on Wednesday.

According to Lookout, the attacks go beyond members of Egyptian and Lebanese civil society, and include targets in the Bahraini and Egyptian governments, as well as targets in the United Arab Emirates, Saudi Arabia, the United Kingdom, and potentially the United States or alumni of American universities.

Lookout concluded that the hackers behind this espionage campaign work for a hack-for-hire vendor with connections to BITTER APT, a hacking group that cybersecurity companies suspect has ties to the Indian government.

Justin Albrecht, principal researcher at Lookout, told TechCrunch that the company behind the campaign may be an offshoot of the Indian hack-for-hire startup Appin, and noted one such company named RebSec as a possible suspect. In 2022 and 2023, Reuters published extensive investigations into Appin and other similar India-based companies, which exposed how these companies are allegedly hired to hack company executives, politicians, military officials, and others.

Techcrunch event This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. San Francisco, CA | REGISTER NOW

Appin apparently later shut down, but Albrecht noted that the discovery of this new hacking campaign shows that the activity “didn’t disappear and they just moved onto smaller companies.”

These groups and their customers get “plausible deniability since they run all the operations and infrastructure.” And for their customers, these hack-for-hire groups are likely cheaper than purchasing commercial spyware, said Albrecht.

Rebsec could not be reached for comment, as the company has deleted its social media accounts and website.

Contact Us Do you have more information about Rebsec Solutions? Or other hack-for-hire companies? From a non-work device, you can contact Lorenzo Franceschi-Bicchierai securely on Signal at +1 917 257 1382, or via Telegram and Keybase @lorenzofb, or Do you have more information about Rebsec Solutions? Or other hack-for-hire companies? From a non-work device, you can contact Lorenzo Franceschi-Bicchierai securely on Signal at +1 917 257 1382, or via Telegram and Keybase @lorenzofb, or email .

⁨Mohammed Al-Maskati⁩, an investigator and director at Access Now’s Digital Security Helpline who worked on these cases, said that “these operations have become cheaper and it’s possible to evade responsibility, especially since we won’t know who the end customer is, and the infrastructure won’t reveal the entity behind it.”

While groups like BITTER may not have the most advanced hacking and spy tools, their tactics can still be highly effective.

In the attacks part of this campaign, the hackers used several different techniques. When targeting iPhone users, the hackers tried to trick targets into giving up their Apple ID credentials in order to then hack into their iCloud backups, which effectively would have given them access to the full content of the targets’ iPhones.

This is “potentially a cheaper alternative to the use of more sophisticated and expensive iOS spyware,” according to Access Now.

When targeting Android users, the hackers used a spyware called ProSpy, masquerading as popular messaging and communications apps like Signal, WhatsApp, and Zoom, as well as ToTok and Botim, two apps that are popular in the Middle East.

In some cases, the hackers tried to trick victims into registering and adding a new device — controlled by the hackers — to their Signal account, a technique that has been popular among various hacking groups, including Russian spies.

A spokesperson for the Indian embassy in Washington, D.C. did not immediately respond to a request for comment.

Tubi, the Fox-owned streaming service, announced Tuesday the launch of its native app within ChatGPT, giving viewers an easier way to discover its large library of more than 300,000 movies and TV episodes.

While competitors like Netflix and Amazon Prime Video have experimented with AI-powered recommendations within their own platforms, Tubi is the first major streaming service to build a dedicated experience directly inside ChatGPT.

To access the integration, users can install the Tubi app from the ChatGPT app store and begin by typing “@Tubi” in a prompt. From there, they can make natural-language requests like “a thriller for girls’ night” or “something funny,” and instantly receive curated recommendations tailored to their preferences, all linked back to titles available on Tubi.

The launch comes as competition across the streaming industry intensifies. With endless entertainment options, discovery has become a challenge for all platforms vying for limited viewer attention. Many streamers have even begun incorporating features inspired by social media platforms to keep users engaged, reflecting broader shifts in how audiences consume content.

Tubi’s move builds on earlier experiments with AI. In 2023, the company introduced “Rabbit AI,” a feature within its mobile app powered by ChatGPT that allowed users to ask specific questions and receive personalized recommendations. However, the tool was discontinued the following year.

The new ChatGPT integration suggests a strategic pivot: Instead of trying to replicate AI experiences in-house, Tubi is now meeting users where they’re already turning for answers. ChatGPT reached 900 million weekly active users in February. Tubi reports more than 100 million monthly active users.

In separate news, Tubi recently launched the “Creatorverse Incubator,” a new initiative aimed at supporting emerging content creators. The program offers promotional backing and potential funding opportunities for original shows that will debut exclusively on the platform.

OpenAI first introduced a way for developers to build apps inside of ChatGPT back in October. Now, dozens of companies have launched integrations, such as Booking.com, Canva, DoorDash, Expedia, Spotify, Figma, and Zillow. SeatGeek was the most recent to launch a native app.

Volkswagen subsidiary MOIA America and Uber have started testing autonomous microbuses in Los Angeles, the companies announced Wednesday, the latest step as they prepare to launch a robotaxi service by late 2026.

A year ago, MOIA America and Uber unveiled an ambitious plan to launch a commercial robotaxi service — using autonomous versions of Volkswagen’s electric ID. Buzz minivan — in multiple U.S. cities over the next decade. Los Angeles is the first city on that list.

Testing will start in the next few weeks with about 10 autonomous ID. Buzz vehicles in Los Angeles, according to the company. The production version of the driverless vehicle seats four people.

Paul DeLong, president of commercialization at MOIA America, called Los Angeles a natural market to introduce the company’s autonomous vehicles for ride experiences, “given its long history of shaping car culture and embracing new mobility technologies.”

Since last year’s announcement, MOIA America and Uber set up a joint facility in Los Angeles for day-to-day fleet operations. The test fleet is small for now, but Volkswagen says it will eventually scale to more than 100 autonomous ID. Buzz vehicles. The vehicles will initially launch with a human safety operator on board. Driverless operations are expected to begin in 2027, the company said.

Sascha Meyer, the chief commercial officer for Volkswagen Autonomous Mobility, said this next step reflects the “strong momentum behind the strategy to bring autonomous mobility into real-world operation.”

The MOIA America name is relatively new branding for Volkswagen’s autonomous vehicle projects in the U.S., known as Volkswagen ADMT until early 2026. But the MOIA brand has been around for a while, since it was first launched by Volkswagen in 2018 at TechCrunch Disrupt London, and is best known in Europe, where it operates a ride-pooling service and tests autonomous vehicle tech in Hamburg, Berlin, Munich, and Oslo. The name change is supposed to reflect the connection between the U.S. and Europe entities.

Techcrunch event This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. This Week Only: Up to $482 savings for Disrupt 2026 Offer ends April 10, 11:59 p.m. PT



Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to secure these savings. San Francisco, CA | REGISTER NOW

MOIA America still must navigate a long regulatory process before it can deploy a commercial robotaxi service — meaning driverless vehicles that charge humans for rides — in California. The company will need permits from the California Department of Motor Vehicles, which regulates autonomous vehicle testing and deployment in the state. It will also need a ride-hailing permit from the California Public Utilities Commission.

Uber, meanwhile, has spread out its bets on autonomous vehicles. The company has partnerships with 25 companies that are applying autonomous vehicle technology to delivery, drones, ride-hailing, and trucking. In the U.S., its most visible partnership is with Waymo. But the company has taken its AV ambitions to other markets, locking in agreements with Chinese companies to launch robotaxis in Europe and the Middle East, as well as startups like U.K.-based Wayve.

Uber recently made a deal with Rivian to buy 10,000 fully autonomous R2 robotaxis ahead of a planned rollout in San Francisco and Miami in 2028. Under the agreement, Uber is making an initial $300 million investment in Rivian.