1. $40B in equity, $55B in concrete, 490% in stock: three opposite bets on the AI silicon chokepoint Nvidia has committed $40 billion to equity stakes in AI companies this year, according to TechCrunch's tally.

2. Lawmakers move to ban AI toys the same week OpenAI hands ChatGPT a panic button Two responses to AI inside private space arrived in the same week, pointing in opposite directions.

3. Tell Claude Code to output HTML and the prototypes start working. One developer's notes hit Hacker News. A Twitter user posting as @trq212 published a piece this week titled "Using Claude Code: The unreasonable effectiveness of HTML.

In Brief

• Moonshot AI raises $2B at $20B valuation as ARR crosses $200M The Beijing-based startup behind Kimi reported its annualized recurring revenue topped $200 million in April, fueled by paid subscriptions and API usage. The round doubles its valuation amid surging demand for open-source Chinese models. TechCrunch

• Mira Murati's deposition details Altman ouster mechanics Witness testimony and trial exhibits in Musk v. Altman document what board members heard in the days before Sam Altman was fired in November 2023. The "not consistently candid" phrase now has specific incidents attached to it. The Verge

• Musk court filings show 2017 attempt to fold OpenAI founders into Tesla Documents from the same trial reveal Musk wanted the OpenAI team to run an AI unit inside Tesla, conditioned on him controlling the for-profit conversion. The founders rejected the terms. Ars Technica

• Mozilla credits Mythos with 271 Firefox vulnerabilities at near-zero false positive rate The browser maker says it has fully committed to AI-assisted bug discovery after running the tool against its codebase. The disclosure gives the first production-scale data point on AI vulnerability scanning accuracy outside the vendor's own benchmarks. Ars Technica

• Google adds source links throughout AI Overviews after publisher backlash Search results will cite sources in several new positions inside generated answers, partially reversing the single-citation design that publishers blamed for traffic collapse. The change affects ad-supported sites whose pages feed the summaries. Ars Technica

• Sony forecasts AI-driven flood of new game releases PlayStation executives told investors that cheaper production tools will increase the volume of titles competing for storefront slots. Sony added that human creative leads must remain in charge of projects, without specifying enforcement mechanisms. Ars Technica

• Tom Steyer proposes California jobs guarantee for AI-displaced workers The gubernatorial candidate's plan would commit the state to providing public employment to workers laid off due to automation. The proposal faces a steep path through the legislature and a tight 2026 primary. Wired

• Google launches screenless $100 Fitbit Air and folds Fitbit app into Google Health The new wearable drops the display in favor of voice and phone-based interaction, with preorders open today. Existing Fitbit users will migrate to the Google Health app, which absorbs sleep, exercise, and metrics tracking. Ars Technica

• Researchers demonstrate remote takeover of internet-connected robot lawn mower Security researchers found exploits letting attackers steer a popular consumer model across property lines without physical access. The disclosure adds to a growing list of yard-equipment vulnerabilities tied to cloud connectivity. Wired

• Paper proposes adaptive execution for robot world-action models Researchers reframe action execution as a verification problem, letting robots stop a planned sequence when imagined futures diverge from observed reality. Current World Action Models execute fixed action chunks blind to mid-rollout drift. Hugging Face

Read the full edition →