Daily ·
Nvidia's Huang resigns; nonprofits sue five AI labs
Nvidia's CEO resigns and Congress closes in on China chip curbs, nonprofits sue five frontier labs, Gemini 4 Hydrogen becomes the default model in AI search, and a Caine Prize investigation clears a writer accused by an AI detector.
🎙 Listen to Skylar
Ten stories from the last 24 hours. The AI boom's most visible executive is gone, the policy response is already moving, and the model releases did not pause for either.
- Jensen Huang resigns as Nvidia CEO — The board confirmed Huang's exit as chief executive, president and director effective immediately, after reports of affairs with a woman US authorities suspect of spying for China and with two employees. CFO Colette Kress is interim CEO. Shares closed down 4.3% after falling 13% at the open, roughly $180 billion off Nvidia's value on the day and about $490 billion for the week. The Guardian · Nov 12
- Nonprofits sue five AI developers to halt frontier development — The Alliance for Secure AI, with the Irreplaceable Education Fund, Public Citizen and the Heat Initiative, filed in San Francisco Superior Court against Anthropic, Google, Meta, OpenAI and SpaceXSI. The suit argues the companies admit they cannot reliably control their systems while handing research and development to AI itself. CBS News · Nov 12
- Congress moves toward tighter China chip export curbs — House Speaker Mike Johnson said there is a "developing consensus" on limiting chip exports to China. Alice Huang, the businesswoman at the centre of the Nvidia scandal, was an executive at Megaspeed, an Nvidia partner the US suspects of helping China obtain restricted chips. Beijing denies she is a spy. CNA · Nov 13
- Gemini 4 Hydrogen is now the default model in AI search — Google's new mid-tier model scores 51 against the larger Argon's 50 on the Artificial Analysis Intelligence Index at default thinking, and reaches 54 at a new "xhigh" tier. It carries a one-million-token input and output window at $1.50 per million input tokens and $7.50 per million output, and lifts Terminal-Bench 4.1 from 18.2% to 57%. The Decoder · Nov 12
- Google and Samsung open preorders for Gemini smart glasses — Nine styles across two eyewear brands: three Warby Parker models from $395 and six Gentle Monster models at $449. The frames run Android XR, weigh about 41 grams, and offer up to nine hours of battery with a case that adds seven charges. PYMNTS · Nov 12
- Fireworks AI breach exposed Notion customers' AI chat transcripts — Fireworks said an attacker compromised its cloud provider to reach a development system, taking customer billing information and some datasets. Notion, which disconnected from Fireworks on November 3, says about 1% of its customers were affected and that chat transcripts from its "Insights" feature were taken. It is Fireworks' second breach in two months. TechCrunch · Nov 12
- ElevenLabs ships Scribe v3 and Scribe v3 Live — ElevenLabs claims about 25% fewer word errors than its predecessor across 102 languages, with real-time speaker diarization available for the first time in the live model. Artificial Analysis puts accuracy near the top of the market, but speed and price mid-pack — Microsoft and xAI run many times faster for roughly half the cost. The Decoder · Nov 12
- Nebius plans $40 billion of 2027 spending, financed by its customers — The neocloud laid out the buildout on its Q3 call, to be funded by customer prepayments and loans against customer contracts. Its Microsoft contract is worth up to $19.4 billion, but a Vineland, New Jersey site is months behind: just over half of that capacity now lands by year-end, with the rest in Q2 2027. TechCrunch · Nov 12
- Caine Prize clears Zimbabwean author accused of using AI — A month-long investigation found Simbarashe Kundizeza's The Blue Fighter, which won the £10,000 prize on October 6, was human-written, after an Atlantic piece reported that the Pangram detector had flagged it as almost entirely machine-generated. The review included interviews and metadata analysis of the author's drafts. The Guardian · Nov 12
- No, AI agents did not discover an unknown Maya site — A Google researcher said a team of 48 AI agents combing NASA lidar data had found an unregistered site in Mexico. Archaeologists noted the site had been documented in an academic journal more than a year earlier. He described the work as a personal side project done without Google compute. The Verge · Nov 12
Try This Next
- 🔧 Run your hardest multi-part question through Gemini 4 Hydrogen at the new "xhigh" thinking tier and compare the answer against Argon.
- 🎥 Listen for how Scribe v3 Live handles crosstalk before you rebuild a voice agent's turn detection.
- 📝 Save the Caine Prize ruling next to your notes on AI detection: a detector score is an accusation, not evidence.
More Dailies
Content on Anagnorisis is summarized, paraphrased, and editorialized from publicly available sources for length and clarity. Original sources are linked where available. All trademarks belong to their respective owners.
