Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡

Here’s what we got for ya today:

Let’s get right into it!

Sign Up | Advertise | Tools

Under oath in a California federal courtroom, Elon Musk confirmed what many in the AI industry had long suspected: xAI used OpenAI's models to help train Grok. Asked point-blank whether xAI had employed distillation — a technique that involves prompting a competitor's model extensively to extract its knowledge and transfer it to a new model — Musk replied "Partly," and added that it was "standard practice to use other AIs to validate your AI." When pressed on whether that meant yes, he did not deny it.

The admission came during week one of Musk's lawsuit against OpenAI, CEO Sam Altman, and co-founder Greg Brockman, who Musk alleges abandoned the lab's nonprofit founding mission in pursuit of profit. The disclosure adds a notable layer of irony: the man suing OpenAI over its alleged misuse of training resources apparently borrowed from OpenAI's own technology to build his rival product. When asked to rank the world's leading AI labs, Musk placed Anthropic first, followed by OpenAI, Google, and Chinese open-source models — describing his own xAI as "a much smaller company with just a few hundred employees."

→ Read More

Families who say AI chatbots contributed to the mental health deterioration or deaths of their children are pressing Congress to act — and fast. In letters to Senate committee chairs obtained by Axios, the parents wrote that "Big Tech deliberately designed their products and platforms to addict, manipulate, exploit, and abuse children and teens." The push is timed to advance the GUARD Act, a bipartisan bill sponsored by Senator Josh Hawley, ahead of its scheduled Senate committee markup.

The GUARD Act would establish a federal framework for AI chatbot safety around minors, including restrictions on manipulative design features and requirements for parental oversight tools. A companion bill, the CHATBOT Act from Senators Ted Cruz and Brian Schatz, would require AI companies to build family accounts that give parents visibility and control over how their children interact with AI. At least six states have already advanced or passed related legislation, but advocates say a federal floor is needed to prevent a patchwork of inconsistent rules from letting companies exploit regulatory gaps.

→ Read More

Meta has acquired Assured Robot Intelligence (ARI), a startup developing foundation models for humanoid robots that can perform physical labor — from household chores to complex, unstructured tasks in dynamic real-world environments. ARI's co-founders, Xiaolong Wang (previously a researcher at Nvidia and associate professor at UC San Diego) and Lerrel Pinto (formerly of NYU, who also co-founded the kid-sized humanoid startup Fauna Robotics before Amazon acquired it last month) will join Meta's Superintelligence Labs division. "We acquired Assured Robot Intelligence, a company at the frontier of robotic intelligence designed to enable robots to understand, predict, and adapt to human behaviors in complex and dynamic environments," a Meta spokesperson told TechCrunch.

The deal reflects a broader sprint across Big Tech to lock down robotics talent before the market matures. Meta has had an internal humanoid hardware team since last year, led by former Cruise CEO Marc Whitten, targeting consumer use cases like home assistance. Whether Meta ever ships a humanoid product, AI researchers increasingly argue that training models in the physical world — where robots learn through direct experience rather than datasets alone — may be the most viable path to AGI. Industry forecasts range from Goldman Sachs' $38 billion projection by 2035 to Morgan Stanley's $5 trillion estimate by 2050, a spread that reflects both the enormous potential and the massive execution uncertainty ahead.

→ Read More

Anthropic is asking investors to lock in funding allocations within the next 48 hours, with the round expected to close within two weeks, according to sources familiar with the matter. The raise is targeting approximately $50 billion at a valuation of around $900 billion — and given the intensity of investor demand, sources say the final figure could exceed even that number. If completed at that price, Anthropic would surpass OpenAI, which raised its landmark $122 billion round at an $852 billion post-money valuation earlier this year.

The round is expected to be Anthropic's last private fundraise before a public offering later this year. Some early backers — particularly those who invested in 2024 or earlier — are reportedly sitting this round out, choosing to wait for liquidity at IPO rather than dilute at a high entry price. The capital will fund Anthropic's escalating compute needs as its annual revenue run rate has reportedly reached $40 billion, ahead of the $30 billion figure the company publicly announced this month. Anthropic raised its previous round in February at a $380 billion valuation — meaning its implied value has more than doubled in under three months.

→ Read More