AI Daily Briefing — 2026-05-15

When Jennifer started a research job in 2023, she ran her new headshot through facial recognition software to see if it would find old porn videos she had made years earlier. It did—but it also uncovered something shocking: one of her videos now featured a different face on her body. This case highlights a neglected side of the deepfake crisis, where adult content creators say AI systems are training on their work, cloning their likenesses, and generating explicit material without consent. Victims face little legal protection or control over their own bodies, according to MIT Technology Review AI.

In a related privacy concern, generative AI chatbots are leaking people’s personal phone numbers. A software developer received unsolicited WhatsApp messages after Google’s Gemini surfaced his number. A university researcher got the chatbot to reveal a colleague’s private cell, and a Reddit user reported a stream of callers seeking lawyers. Experts believe these breaches stem from personally identifiable information in training data, and victims have few options to stop the spread.

Meanwhile, the Tesla Semi is finally rolling off production lines nearly a decade after its unveiling. The electric truck, which reportedly travels up to 480 miles per charge, could boost battery-powered freight and reduce pollution from semitrucks. Despite high prices and charging challenges, Tesla’s model costs far less than competitors, offering a vital step forward for electric trucking, reports MIT Technology Review AI.

A startup named Varda Space Industries is pushing forward with plans to manufacture pharmaceuticals in orbit. The company has entered into an agreement with United Therapeutics to explore how microgravity affects drug crystallization, potentially leading to improved medications with novel properties. While the concept may seem like science fiction, decreasing launch expenses and the advent of reusable rockets are making orbital production increasingly viable. Varda views this collaboration as a significant milestone toward creating products in space for terrestrial use.

In other space news, NASA is developing its first interplanetary spacecraft powered by a nuclear reactor, with a target launch by the end of 2028. The mission aims to travel to Mars, potentially marking a transformative moment for spaceflight and giving the United States a competitive advantage over China. However, details about the project remain limited. Experts consulted by MIT Technology Review AI shed light on how this nuclear propulsion system might function, though many aspects are still under wraps.

The push for orbital manufacturing and nuclear space travel reflects a broader trend of commercial and governmental entities expanding humanity's presence beyond Earth. As technology advances, the line between science fiction and practical innovation continues to blur.

Cerebras Systems made a stunning debut on the public market Thursday, raising $5.5 billion in what is being hailed as the first major tech IPO of 2026. The AI chipmaker priced its shares at $185 Wednesday evening, well above its initial range of $115 to $125, after increasing the offering to 30 million shares. When trading opened, the stock soared to $385—a 108% jump—before settling above $330 by midday, driven by heavy retail investor demand.

At the IPO price, Cerebras entered trading with a fully-diluted valuation of $56.4 billion. Co-founder and CEO Andrew Feldman’s stake is now worth nearly $1.9 billion, while co-founder and CTO Sean Lie holds shares valued at about $1 billion. If the stock holds above $300, both founders will see their fortunes rise even further by the close of trading.

The path to this milestone was far from smooth. Cerebras, which designed a massive chip from scratch to compete with Nvidia, first filed for an IPO in 2024 but was stalled by a CFIUS review over a large investment from Abu Dhabi’s Group 42. Investor skepticism also lingered, as Group 42 accounted for nearly all of Cerebras’s revenue at the time. The company shelved those plans until April, when it reported $510 million in 2025 revenue—up 76% year-over-year—and swung to a $237.8 million profit from a nearly $500 million loss the prior year.

Now a major player in AI inference chips, Cerebras counts OpenAI, G42, Saudi Arabia’s Mohamed bin Zayed University of Artificial Intelligence, and Amazon Web Services among its customers. The successful IPO marks a dramatic turnaround for the company and signals strong market appetite for AI hardware investments.

The clock is ticking for early-stage founders aiming for a spot at TechCrunch Disrupt 2026. Applications for the Startup Battlefield 200 close on May 27, leaving just two weeks for startups to secure a chance at $100,000 in equity-free funding, global media exposure, and direct access to top venture capitalists. TechCrunch AI reports that this year’s cohort will feature 200 of the most promising pre-Series A companies, with only 20 finalists pitching live on the main stage.

Selected startups receive a free exhibit table for all three days of Disrupt, four all-access passes, and a featured profile in the event app. They also gain exclusive access to founder masterclasses, press list opportunities, and direct feedback from leading VCs. The program has previously accelerated companies like Dropbox, Discord, and Fitbit, making it a high-ROI launchpad for category-defining ideas.

Organizers emphasize that early applicants gain an edge, with more time to prepare and stand out to TechCrunch editors. The strongest startups are already entering the arena, and late submissions risk getting lost in the noise. Founders with a functional MVP, a clear demo, and strong market traction are encouraged to apply across all industries globally.

With thousands vying for just 200 slots, the competition is fierce. Nominations are still open for those who know a deserving startup. The message is clear: the founders who break through aren’t waiting until the final hour. Apply by May 27 to stake your claim on the Disrupt stage.

Notion, the popular productivity software company, is making a major shift into the age of AI agents. During a livestreamed event on Wednesday, the company unveiled a new developer platform designed to turn its workspace into a central hub where both people and AI agents can collaborate. The platform includes an orchestration layer that coordinates AI work across multiple tools and data sources, allowing teams to build automated, multistep workflows that pull information from any database.

Since launching custom AI agents in February, Notion users have created over one million of these digital teammates to handle repetitive tasks like answering questions and compiling updates. However, those agents were limited because they couldn't connect to external data or use custom logic. The new platform addresses these gaps by introducing Workers, a cloud-based environment where teams can deploy their own code in a secure sandbox. This enables syncing data from external sources like Salesforce or Postgres directly into Notion databases without relying on third-party infrastructure.

Notion is also opening its system to external AI agents. Users can now chat with agents from partners like Claude Code, Cursor, and Codex, assign them tasks, and track their progress as if they were native Notion agents. An External Agent API allows companies to connect their own custom-built agents as well. Notion co-founder and CEO Ivan Zhao acknowledged the company's past limitations for developers but emphasized that things are changing, positioning Notion as more than just a note-taking app with AI features.