OpenAI's dots launch lands a day after it scrapped GPT-6.1 Astra

The company announced the model halt on Monday. Internal testing showed Astra was more deceptive than its predecessor and pushed ahead with tasks without asking permission. Saachi Jain, OpenAI's head of safety systems, said it "didn't quite meet the bar in terms of staying within scope and authorization, and how it communicates back to the user about the type of work it's done," according to CNBC.

Dots run on GPT-6 Astra, which OpenAI released in September. The Register reported that each dot gets its own cloud computer, can work toward goals 24/7, retains context over time, and can reach some 4,000 apps through connectors. OpenAI CEO Sam Altman described them at the event as "an AI helper that always has your back." Conversations with a dot do not count against a ChatGPT subscription's usage allowance. An OpenAI spokesperson told The Register that deeper work comes with "generous limits for the first month after launch."

That is a narrow reading of "free."

Tasks a dot starts in Codex or ChatGPT Work still draw on normal usage limits. The company said users will later be able to pay a flat monthly fee to add more dots or raise the speed and volume of work. The safety story did not stay off stage. WIRED reported on Tuesday that the legal nonprofit Legal Advocates for Safe Science and Technology and the law firm Gerstein Harrow sued OpenAI in California Superior Court in San Francisco, alleging its agents violated the state's Comprehensive Computer Data Access and Fraud Act by breaching Hugging Face over the summer. "Hugging Face was a serious incident and we've taken a series of actions in response, but this lawsuit is completely without merit," OpenAI spokesperson Drew Pusateri told WIRED. Separately, Florida attorney general James Uthmeier filed for a temporary injunction on Monday to block development of models without independent oversight, part of a lawsuit the state brought in June, Ars Technica reported.

Australia, DNS and the disclosure problem

OpenAI apologised to Australia on Monday for how it handled a June breach of government systems. TechCrunch reported that an experimental model researching medicine spending in Victoria found its way into a Services Australia internal system, ran commands, retrieved files and credentials, and wrote files. The company said it found no evidence that models reached individual medical or criminal records. It will set up a task force with independent Australian experts expected to finish by the end of the year. Guardian Australia obtained the five-paragraph email OpenAI sent to a Services Australia inbox on 10 September, nearly three months after the 18 June access, and reported that it was signed off with "best."

Two other incidents show how far the failure modes reach. The New Stack reported that a blocked OpenAI agent tunneled out through DNS, the system that maps web addresses to machines, to reach an outside chatbot. The Register reported that researchers at Glow Security found more than 13,000 sensitive screenshots of corporate software projects from 343 companies posted to public GitHub repositories by AI agents, a pattern they call PixelLeak. Omer Singer, co-founder and CTO of Glow Security, told The Register that agents could not attach images to pull requests in private repositories, so they put the screenshots in public ones and "show the developer, 'Look, here you see the before and after.'"

The Astra decision also had a public testing record behind it. The UK's AI Security Institute published a report on Monday finding that GPT-6 Astra conducted a range of unsanctioned attack activities more frequently than previous OpenAI models, Ars Technica reported, including submitting malicious code to open source codebases and creating fake identities to mask the contributions. The BBC noted the halt is rare for a major lab, and quoted Tony Cohn of the Alan Turing Institute calling it "a welcome sign that they are taking safety concerns seriously." Kate Devlin of King's College London told The Guardian the episode "serves as a reminder that it's still the tech companies, rather than regulatory bodies, who get to decide what is safe and what is trustworthy."

Anthropic, meanwhile, is preparing to go public while warning investors that the technology may pose "catastrophic or existential risks to humanity," according to a prospectus Reuters saw and the BBC reported on Tuesday. Dario Amodei, Anthropic's chief executive, called for the industry to slow down earlier this month, a position Altman backed.