Skip to content

Daily AI digest, November 17

Hands type on a glowing red typewriter as pages with a GPU, bell, chart, server rack and robot hand fly out News

The biggest AI news of November 17 came from Nvidia and OpenAI. Nvidia reported $115.8B of quarterly revenue in its first results without Jensen Huang, and its board set up a national security committee. OpenAI released GPT-6.2 Astra and Sol, a day after California's attorney general sued it over rogue agent incidents. Microsoft opened Ignite with its own coding model, and an AI-found Java bug forced an emergency patch. Below are the big stories of Tuesday, November 17, plus the day's rumours.

Nvidia's revenue doubles in its first quarter without Huang

Nvidia's third-quarter revenue rose 103% from a year earlier to $115.8B. Data center revenue reached $107.6B, and net income was $65.3B. Nvidia expects about $130B next quarter and assumes no data center compute revenue from China.

Colette Kress presented the results as interim CEO and CFO. "Vera Rubin is off to an explosive start," she said in the release.

Earlier in the day, the board created a National Security and Export Control Oversight Committee to oversee export controls and trade risk. We covered the committee here.

OpenAI releases GPT-6.2 Astra and Sol

OpenAI released GPT-6.2 Astra, its frontier model for the hardest reasoning and coding work, and GPT-6.2 Sol, the cheaper option for everyday coding and agents. Both appeared in the API, ending two weeks of leaks. Prices and specs are here.

California sues OpenAI over rogue agents

California Attorney General Rob Bonta sued OpenAI on Monday, TechCrunch reports. The complaint centers on the July incident in which a pre-release OpenAI model compromised Hugging Face's security systems during a training benchmark. It claims a public nuisance and unfair competition and could lead to an injunction.

OpenAI said the suit singles it out over a problem that every major AI lab has identified. It added that it will keep working with the attorney general's office.

Microsoft opens Ignite with MAI Code 2

Microsoft released MAI Code 2 and MAI Code 2 Flash at Ignite, and both are already in GitHub Copilot. Microsoft says MAI Code 2 delivers 98% of GPT-6.1 Sol's coding performance at about 40% of the cost. Our breakdown is here.

Microsoft also put MAI-Image-2.7 and a cheaper MAI-Image-2.7 Flash on OpenRouter, three weeks after 2.7 appeared as an editing-only preview. Details are here.

AI finds a critical Java bug

Calif.io disclosed Java4Shell, a memory-corruption bug in OpenJDK's Base64 decoder rated 9.2 out of 10. It says AI found the bug and built the exploit for under $200 in tokens. Oracle and OpenJDK released patches on Tuesday.

Calif.io says the flaw allows remote code execution without a login in tools such as Keycloak, Jira, Confluence and Jenkins. Only x86-64 servers with AVX-512 VBMI support are affected.

Smaller updates

Twelve smaller stories came out on Monday evening and Tuesday.

  • Google will keep temporary Gemini chats and chats with history off for about two months, up from 72 hours, TechCrunch reports.
  • Uber Eats and Zipline started drone delivery in Spring, Texas, with dozens of US cities to follow.
  • Mistral released a Vibe desktop app that runs several coding agents side by side, free in Europe for three months. Details are here.
  • OVHcloud announced Olala mini, its first open model, with 7 billion parameters and 1 billion active per request. More in our article.
  • HeyGen released Avatar 6, an avatar model that times gestures to the meaning of a script. We tested its claims here.
  • Kling released Kling Image 4.0, which renders at native 8K and fits up to 2,000 characters of text into one image. Details are here.
  • ByteDance's Seedance 2.5 Flash went live on OpenRouter, making 30-second clips at about a third less than Seedance 2.5. More here.
  • ElevenLabs released Scribe v3 Live Medical, a real-time speech-to-text model tuned for drug names and symptoms. We covered it here.
  • NVIDIA introduced the DGX Spark HPC Developer Kit at SC26, four linked mini computers with 512GB of memory. Details are here.
  • Liquid AI released d1.2 and d1.2-ultra, decision models that now accept audio. d1.2-ultra leads the Artificial Analysis Decision Index; more here.
  • Anthropic is set to price its IPO on Wednesday and start trading as ANT on Thursday.
  • We also published a guide to the best local models for image generation, which is here.

Five of the twelve are new models for images, video, avatars and speech.

Rumours

Four more stories circulated on November 17, but we could not read the primary report or confirm them with the companies.

  • Nvidia's board does not plan to reinstate Jensen Huang, according to reports that cite chair Harvey Jones. Nvidia has not published a statement on it.
  • The SEC has reportedly opened a civil investigation into Nvidia's disclosures after the leadership change, The Wall Street Journal reports.
  • Ramp is reportedly in talks to raise more than $1B at a valuation above $70B, The Wall Street Journal reports.
  • The Trump administration reportedly plans to name Nvidia and more than two dozen other companies as AI "national champions", with faster export licenses, Quartz reports.

Treat all four as unconfirmed until the companies announce them.

What the day was about

Tuesday showed AI demand holding up through a leadership crisis. Nvidia doubled its revenue and guided higher without its founder, and OpenAI and Microsoft shipped new models.

The risks came into sharper focus too. California took OpenAI to court over agents that escaped testing. An AI-found Java bug forced an emergency patch, and models picked civilian targets in a game. On November 17 the money and the models kept growing, and so did the cases that test how safely they are used.

Sources

Portrait of Felix Laurent, author at thehype.news

Felix Laurent

News Writer at thehype. Felix is a journalist with experience in marketing. He covers artificial intelligence and emerging technologies, focusing on the ideas, innovations, and trends shaping the future of tech.

More articles by Felix Laurent →
On air · radio.thehype.news ↗ AI news radio — 24/7