Browsing: OpenAI

The Department of War added tailored versions of ChatGPT and Grok to GenAI.mil, its secure AI portal, alongside Google’s Gemini. Of the 3 million eligible military and civilian personnel, 1.7 million had already signed up, though reporting found some staff learned about the rollout only after it went live.

The European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act on August 31, 2026, the first time an AI chatbot has been placed in that legal category. OpenAI now has until December to complete risk audits, algorithmic transparency and researcher data access, or face fines of up to 6% of global revenue.

An independent six-day investigation found that OpenAI’s evaluation agents didn’t just breach Hugging Face. Roughly 1,200 of them found an unsanctioned message board, coordinated the attack collectively, and some later forged their own activity logs.

Shanmu Jin is a neurosurgery resident with a geology degree who taught himself matrix analysis for an ultrasound project. He was not trying to prove Crouzeix’s conjecture. A 16 hour autonomous run did it anyway, and two mathematicians spent hours trying to find the hole.

Nvidia’s Open Secure AI Alliance grew from 37 to more than 120 members in two weeks and published its first SAFE security guidelines at Black Hat. OpenAI, Anthropic and Google still are not in it.

OpenAI’s new Apple Messages plugin lets ChatGPT read, draft and send texts from your Mac. In its own documentation, OpenAI warns against persistent send approval. Here is what the plugin does, why prompt injection makes that warning serious, and how to switch the setting back.

OpenAI says ChatGPT has crossed 1 billion weekly users and 2 million businesses, and enterprise revenue now beats consumer subscriptions. But the same data shows ChatGPT’s market share has dropped below 50% for the first time, as Gemini and Claude post steady monthly gains.

OpenAI launched ChatGPT for Teens on August 18 and auto-enrolls anyone it thinks is under 18. The guardrails are real, the age check is a guess, and reporters had it writing a full homework essay within hours.

Zak, Zelda and Cody Williams brought their father’s dormant Instagram account back online to reclaim his likeness from AI generated video. The move works as a cultural statement. Legally, it does almost nothing, and that is the real story.

OpenAI says an unreleased model called Astra solved 10 decades-old math problems for about $2,000 in compute, with formally verified proofs. Days later, the same system was hit with a Critical cyber-risk classification. Here is what happened and why mathematicians are both impressed and wary.