WireGuard, a prominent software project and VPN that forms the backbone of many security applications including Mullvad and others, has encountered a blockade in a significant segment of its Microsoft developer account, rendering it unable to release software updates to Windows users.

Jason Donenfeld, the inventor of the open source WireGuard VPN software, informed TechCrunch that he has been barred from accessing his Microsoft developer account, consequently hindering his ability to sign drivers or send updates for WireGuard to Windows users, which are essential for its operation. Donenfeld mentioned in a post on X on Wednesday that the account termination has prevented a WireGuard update from being released.

This marks the second occurrence of a prominent and widely utilized open source project being denied access to its users due to what appears to be an abrupt account termination by Microsoft, with the well-known encryption tool VeraCrypt experiencing a similar situation. Both developers claimed that Microsoft locked them out of their accounts without prior notification. 

In the situation concerning VeraCrypt, which is relied upon by hundreds of thousands of users to secure files and operating systems, its developer Mounir Idrassi conveyed to TechCrunch that being barred from his account means he cannot update the software in time for an important certificate authority expiration, which he stated might prevent certain users from starting up their systems.

Donenfeld, the WireGuard developer, told TechCrunch via email: “If there were an urgent vulnerability to fix at this moment — there isn’t! I’m just speaking hypothetically — then users would be completely vulnerable.”

WireGuard is a globally utilized open source VPN software that connects devices via the internet. Its code is well-regarded for its simplicity and security, serving as the basis for numerous VPN implementations and commercial services that depend on its code, such as Proton and Tailscale.

Donenfeld communicated to TechCrunch in an email that he has dedicated the last few weeks to modernizing WireGuard’s Windows code and was prepared to submit an updated copy to Microsoft for review prior to its release to users, but encountered an “access restricted” error when attempting to log into the developer section of his Microsoft account.

Despite completing the process to authenticate his driver’s license or passport with Microsoft (the third-party Microsoft employs for verification stated he was “verified”), Donenfeld reported that his access remained suspended.

Donenfeld found a page on Microsoft’s website indicating that the company had been conducting “mandatory account verification for all partners in the Windows Hardware Program who have not completed account verification since April 2024,” though the verification initiative has since concluded.

Microsoft’s Windows Hardware Program permits developers like Donenfeld and VeraCrypt’s Idrassi to “deploy hardware and device drivers for Windows PCs and other devices.” The ability to craft and distribute drivers for Windows users is confined to recognized and vetted developers, as drivers can provide extensive access to an operating system and its data, which are often exploited by hackers.

This account verification process necessitated that developers upload government-issued ID before they could publish potentially sensitive code to the larger Windows user community.

“Microsoft never sent me any notification whatsoever about this. I’ve checked every inbox, every spam folder, and every mail log, and found absolutely nothing,” Donenfeld stated.

The Windows Hardware Program’s verification initiative has “now concluded” and developers who did not upload their documents had their accounts “suspended,” the page indicates, which means these accounts can no longer provide updates.

Donenfeld mentioned that he was directed to Microsoft’s executive support team, which manages customer service and account inquiries for high-profile clients, confirming that his appeal was acknowledged, but they may take up to 60 days for assessment.

By late Wednesday, there was a hint of optimism in Donenfeld’s scenario. He informed TechCrunch that he has finally made contact with Microsoft and is hopeful for a swift resolution.

Microsoft did not provide immediate comments when contacted by TechCrunch.

Donenfeld and Idrassi are not isolated in this situation, as the account lockout problems are impacting others too.

Windscribe, a developer of VPN and other consumer privacy solutions, announced in a post on X that it had also faced lockout from its Partner Center account. The company stated that it maintained a verified account for over eight years to sign its drivers.

“We’ve been attempting to resolve this for more than a month, with no progress. Support is nonexistent,” Windscribe expressed in its post. “Is there anyone who knows a competent person at Microsoft who can assist?”

The passing of a duck in the Mueller Lake area of Austin, Texas, has led neighbors to voice their worries regarding the presence of autonomous vehicles there.

Although it is common for humans to unintentionally harm animals with their vehicles, this incident has cast a shadow over the emerging technology. Local news began covering the duck incident after a community member shared in a Mueller neighborhood Facebook group that an Avride autonomous vehicle (with a human safety operator at the controls) struck and killed a duck without stopping. “It didn’t pause or hesitate whatsoever, just plowed through,” the post, highlighted by KXAN, states.

The residents’ bond with this particular duck, which was nesting in a flower pot outside a nearby Italian restaurant, has fueled the outrage and distrust toward autonomous vehicle technology. For those worried about the fate of the duck’s eggs, local residents have placed them in an incubator, as reported by Axios’ Austin.

An Avride representative confirmed to TechCrunch that the vehicle was operating in autonomous mode during the incident. Avride has not fully halted its public road testing. Nevertheless, the company has modified its operational area by omitting specific streets in the vicinity of the lake where the duck incident took place, according to spokesperson Yulia Shveyko.

The resident further alleged in their post that the vehicle did not stop at a stop sign. Avride informed TechCrunch that it found no evidence supporting this assertion. The vehicle made full and appropriate stops at all relevant stop signs.

Shveyko mentioned that the team has analyzed the vehicle’s data and behavior, including multiple simulations of the incident. Avride is currently assessing potential enhancements to the technology to help prevent similar occurrences in the future, she stated. Importantly, this involves conducting a series of controlled experiments in simulations to ensure that any modifications do not adversely affect the vehicle’s safety performance in other situations.

Avride is not the sole company testing or deploying autonomous vehicles commercially in the area. Zoox has also been conducting tests in the city. Additionally, Tesla and Waymo, in collaboration with Uber, provide a commercial robotaxi service in certain parts of Austin.

Is Poke an OpenClaw accessible to all of us? That’s the concept being introduced by a new startup that provides an AI agent available through iMessage, SMS, Telegram, and, in select regions, WhatsApp.

The AI agent Poke was publicly released in March, enabling users to engage with a personal assistant that can execute tasks on their behalf using a familiar interface. At present, Poke can assist with daily requirements such as planning your day, managing your calendar, monitoring your health and fitness, controlling your smart home, editing images, and much more, all through text messaging.

While you might still engage with a general-purpose AI chatbot like ChatGPT or Claude for inquiries or research, you would turn to Poke when you need to accomplish something swiftly, or wish to automate a task for efficiency.

For example, you could request Poke to notify you of specific emails (such as those from family or your supervisor), or remind you in the morning if you should carry an umbrella. It could assist you in monitoring your health and fitness objectives, or inform you about the latest game scores. Poke could send reminders for daily medications or catch you up on current events, and more, as users can create their own automations in simple text and share them with others.

Supported by Spark Capital, General Catalyst, and various angel investors, the 10-person startup has recently secured an additional $10 million, adding to last year’s $15 million seed funding. It is now valued at $300 million, after investment.

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

This tool arrives amidst a surge in demand for agentic AI systems, prompting OpenAI to acquire OpenClaw’s developer, while Nvidia CEO Jensen Huang cautioned that every enterprise should develop its own OpenClaw strategy when unveiling Nvidia’s enterprise-grade alternative.

However, for those less tech-savvy, the idea of installing software via command line, handling dependencies, and resolving errors can seem overwhelming. Additionally, systems like OpenClaw pose security risks due to their extensive system access.

Consequently, many find that OpenClaw and similar agentic systems appear inaccessible. The team behind Poke aims to alter that perception.

Marvin von Hagen, co-founder of The Interaction Company of California, the Palo Alto-based startup behind the AI agent, informs TechCrunch that Poke originated from observing how beta testers interacted with the company’s previous product, an AI assistant designed for email, crafted around a year prior.

“What we observed was that people desired to utilize Poke for numerous tasks… Even though it was initially intended for email only, users began requesting Poke to remind them to take their medications. They inquired about sports scores — ‘Hey Poke, tell me every morning if I need a jacket or not,’” von Hagen clarifies. “At that moment, we lacked many of these features, but we recognized the need to evolve into a more versatile tool rapidly, as users appreciated the personality and relatability of it.”

The team then shifted its focus to enhancing Poke’s utility, proactivity, and personable qualities.

In contrast to OpenClaw, starting with Poke is straightforward. Simply head to Poke.com, click “Get Started,” and input your phone number. There’s no application to download since the assistant functions through text messages.

Internally, Poke selects the AI model that is best suited for the task, whether it is from one of the major AI providers or an open-source model.

“I believe this is one of our primary advantages moving forward: nearly all of our rivals are large tech firms and labs tied to specific providers. For instance, Meta AI will only utilize Meta’s models, and ChatGPT will solely use OpenAI’s models,” von Hagen emphasizes.

To function over messaging platforms like iMessage, Poke also utilizes Linq, a solution that facilitates AI assistants to operate within messaging applications. The app can also be accessed through SMS and Telegram, though WhatsApp support is limited at this time due to Meta prohibiting other general-purpose chatbots since last autumn.

However, this may change. Regulators from the EU, Italy, and Brazil have initiated antitrust investigations to counter this decision, which has prompted Poke’s return to Brazil. It is hoped this will enable Poke to function on WhatsApp in the EU once Meta reduces its fees. (Meta has faced criticism regarding the exorbitant fees it imposes — von Hagen describes it as a form of “malicious compliance” that he believes will be resolved soon.)

Upon launch, Poke provides several “recipes” — these are pre-configured tools that assist in automating different areas of your life or work. These recipes cover aspects such as health and wellness, productivity, finance, scheduling, travel, home, education, email, community, and for those with technical skills, developer tools. Installing them is as easy as clicking a button followed by a customary authorization process, if necessary.

These recipes are crafted to integrate seamlessly with familiar apps and services like Gmail, Google Calendar, Outlook, Notion, Linear, Granola, among others. There are also health and fitness “recipes” compatible with Strava, Withings, Oura, Fitbit, etc., along with options for smart home devices from brands like Philips Hue and Sonos.

Developers using Poke can automate workflows using integrations with tools such as PostHog, Webflow, Supabase, Vercel, Devin, Sentry, GitHub, Cursor Cloud Agents, and others.

Poke’s security framework is multi-faceted, incorporating regular penetration testing, security audits, a variety of tools, and permission limitations for both agents and human operatives. By default, the team cannot access anything within the tokens unless the user voluntarily opts to grant access to a log file or analytics through settings to share this data. (TechCrunch has not conducted an independent security audit, for clarity.)

In recent weeks, Poke users have generated thousands of additional recipes and automations, which the company intends to incorporate into its recipes directory for easier discovery soon. They are also motivating creators to develop these shareable recipes by offering to compensate anywhere from 10 cents to a dollar (depending on location) for each user registering for Poke via the recipe.

Utilizing Poke is surprisingly economical: it’s free to begin, followed by flexible pricing. During the beta phase, users actually negotiated the monthly fee with the AI agent, which varied between $10-$30 — as Poke indicated in response to this inquiry.

Von Hagen mentions that, currently, pricing is contingent on how the AI agent is utilized. If you are requesting information that doesn’t necessitate real-time data, you could likely use Poke for free. Poke incurs costs from real-time inference, such as automations that trigger with every incoming email or during real-time flight check-ins. To establish pricing, the company has guided Poke on the costliness of various tasks, enabling it to provide tailored pricing.

While the company has successfully optimized Poke’s efficiency to minimize expenses, the primary focus at this time isn’t on profitability, notes von Hagen.

“Our primary goal is not to make profits, but to achieve growth. We aspire to create a product for one billion individuals, with monetization being a secondary concern,” he states. “In the upcoming weeks and months, our aim is to integrate Poke into daily life.” To accomplish this, they will rely on creators and influencers to demonstrate how they utilize 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, has not disclosed the number of customers who have joined, only mentioning that the count has multiplied by 10 over the last few months. (However, Poke was noticed at the top of Vercel’s AI Gateway leaderboard, which is noteworthy.)

Alongside its main institutional backers, Spark Capital and General Catalyst, the startup has garnered attention from several angel investors, including John and Patrick Collison (founders of Stripe), Jake and Logan Paul, Logan Kilpatrick from DeepMind, Joanne Jang of OpenAI, and Scott Wu and Walden Yan (founders of Cognition).

Moreover, it has involved 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, among others.

On Wednesday, Canva revealed its double acquisition of Simtheory, a platform for AI collaboration and agent management, alongside Ortto, a company specializing in customer data and marketing automation. The financial aspects of these acquisitions have not been revealed. According to Canva, these moves are part of its ongoing commitment to invest in AI and marketing infrastructure. 

Chris and Mike Sharkey, the founders of both companies, previously established the vacation rental service Stayz, which was bought by Fairfax Media. The duo will take on leadership positions within Canva, focusing on the company’s AI and marketing technology teams.

Canva states that these acquisitions bolster its strengths in agentic AI, data infrastructure, marketing automation, and customer interaction, allowing Canva to transition from a design tool to a comprehensive system where teams can execute all their tasks from inception to completion. These additions enhance Canva’s functionality throughout the entire workflow, from initial concepts to scaling and evaluating campaigns.

The platform by Simtheory aids teams in employing AI to create assistants that grasp their business, integrate across various tools, and manage actual tasks. Simtheory allows teams to implement the latest models for a variety of applications and establish agentic workflows customized to their specifications.

Conversely, Ortto merges a customer data platform with marketing automation, enabling teams to craft and oversee journeys via email, SMS, push notifications, in-app messaging, forms, and surveys all within a single system. Canva asserts that its event-driven architecture and no-code integrations facilitate the seamless connection and real-time activation of data. The platform serves over 11,000 clients across 190 nations.

“Simtheory propels our transformation from a design platform featuring AI tools to an AI platform with design and productivity tools at its foundation,” stated Canva co-founder and COO Cliff Obrecht in an official announcement. “Simultaneously, Ortto enhances our capability to drive the entire marketing and content lifecycle through Canva Grow, from strategizing and creating to publishing and optimizing across all channels.”

Canva Grow serves as the company’s tool for creating assets and measuring performance.

These recent acquisitions arise as Canva has been actively pursuing growth through acquisitions. Just two weeks ago, the firm acquired the digital outdoor advertising startup Doohly. Six weeks prior, Canva announced the dual acquisition of Cavalry, which specializes in animation, and MangoAI, focused on enhancing ad performance.  Earlier, in January 2025, Canva acquired the marketing intelligence startup MagicBrief.

Canva concluded 2025 with an annualized revenue of $4 billion alongside over 265 million users and 31 million paid subscribers. The company also experienced a 20% rise in monthly active users.

AWS CEO Matt Garman stated that Amazon’s recent $50 billion investment in OpenAI, following its longstanding partnership that included $8 billion in Anthropic, reflects the type of conflict of interest the cloud behemoth is accustomed to managing.

Garman has been with Amazon since he interned there during business school in 2005, prior to the debut of AWS in 2006, he shared with the HumanX conference audience happening this week in San Francisco.

When questioned about the intrinsic conflict of collaborating closely with two AI model firms that are intense (and, arguably, sometimes trivial) competitors, he claimed it’s not an issue. Since AWS frequently competes with its partners, it possesses ample direct experience with such rivalry, he elaborated.

During AWS’s initial years, it recognized it couldn’t create every cloud solution independently, so the division allied with others.

“We also understood that we might compete with our partners, as technology is intertwined,” Garman reflected. “Thus, for a considerable time, we’ve developed the capability to market alongside our partners,” he continued. “However, we might even have first-party products that rival them, and that’s acceptable, and we’ve assured them that we won’t give ourselves an undue competitive edge.”

Nowadays, the public is familiar with Amazon competing against those who offer services on its cloud. Even one of AWS’s major competitors, Oracle, provides its database and additional services on AWS. However, this was a groundbreaking concept back in 2006, when tech partners took care not to compete with those that contributed to their success.

Nonetheless, Amazon is hardly pioneering the abandonment of investor loyalty and conflict-of-interest principles in the cutthroat, profit-driven realm of AI. When Anthropic revealed its latest $30 billion funding round in February, it included at least twelve investors who were also supporting OpenAI, including OpenAI’s primary cloud partner, Microsoft.

For AWS, investing significantly in OpenAI to access its model for its clientele (and as a technology development ally) was nearly a matter of survival. Both models were already accessible on Microsoft’s cloud, AWS’s most significant competitor.

The cloud titans are also striving to maintain visibility by providing AI model-routing services. These services enable their clients to automatically deploy different models for various functions, aiming to enhance performance and cut costs. As Garman illustrated, one model may be optimal for planning, another for reasoning, and a more economical model for simpler tasks, like code completion. “I believe that is where the future will lead,” Garman remarked.

That is also how Amazon, and Microsoft for that matter, will integrate their proprietary models into use — that recurring competition-with-your-partners scenario.

All’s fair in love and AI these days.

Amazon is set to discontinue support for Kindle devices launched before 2012 — severing their connection to the Kindle Store and rendering them nearly unable to download any new content. The cutoff date has been established for May 20, 2026.

This week, Amazon informed customers via email that it will cease support for these devices, a message corroborated by Amazon and shared on social media and Reddit through customer screenshots. As stated in the email, Kindle e-readers and tablets introduced in 2012 or earlier will lose Kindle Store functionalities past the deadline, meaning affected devices will no longer be able to purchase, borrow, or download new books directly.

Post May 20, users will be limited to reading content that has already been downloaded on their devices. An older device that is deregistered or restored to its original factory settings cannot be re-registered after May 20.

Amazon’s list of models affected 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 for as long as 18 years — but technology has advanced significantly during this period, and these devices will not receive further support,” said Jesse Carr, an Amazon representative, in a statement to TechCrunch. “We are reaching out to those who are still actively using them and providing promotions to assist with the transition to newer devices.”

Users may continue to access their libraries via newer devices, the Kindle mobile app, or Kindle for Web, according to the Amazon spokesperson.

Many long-term users expressed their dissatisfaction with the decision to terminate support for older Kindles, particularly those still in excellent condition that have not experienced any performance decline. On Reddit, several users noted their ongoing reliance on these older devices and saw no justification for upgrading. For some, it felt like another instance of tech companies discontinuing support for functional products, prompting customers to upgrade and contributing to the escalation of electronic waste. A parallel is drawn with Google, which ceased support for its older Nest thermostats in 2025. According to the United Nations Institute for Training and Research (UNITAR), global e-waste is projected to reach 82 million tonnes by 2030, marking a 32% increase from 2022.

One Reddit user commented, “I’ve owned my Kindle for years, but it continues to function flawlessly and serves my needs well. How wasteful is it to render a product nearly unusable just to compel people to purchase a newer model?”

For some, this change suggests it may be time to consider alternative e-ink readers, such as the Boox Palma or Vivlio e-reader.

The choice to essentially phase out older Kindle devices doesn’t come as a surprise, as Amazon has been progressively phasing out support for its legacy devices. In 2016, the company mandated that Kindle e-reader models from 2012 and earlier install updates merely to retain access to the Kindle Store, an initial indication that long-term support for these devices would eventually be phased out.