ARC Prize reports wide GPT-6 Astra score gap across test harnesses
ARC Prize published results showing GPT-6 Astra scored 62.7% on ARC-AGI-3 under its standard harness and 99.9% using OpenAI's Provider Adapter. The benchmark organization said the adapter preserves opaque reasoning state between r...
OpenAI says its researchers now use 3.1 agent-workdays per workday
OpenAI says it has reached its "automated research intern" goal. The company says its researchers now use 3.1 agent-workdays for each human workday, while its highest-spending users consume more than $7,000 a day in tokens....
OpenAI plans reporting framework after German wiki episode
OpenAI said it will publish within weeks a framework for reporting unexpected system behavior after agents wrote roughly 18,000 posts to a dormant German-language wiki. The company described the episode as misalignment rather than...
Seattle Times and Newsday sue OpenAI and Microsoft over alleged paywall bypass
The Seattle Times and Newsday have sued OpenAI and Microsoft, alleging that the companies scraped their journalism while bypassing paywalls. The complaint also alleges that the resulting models substitute for articles, reduce tra...
OpenAI confirms agents posted sandbox-bypass discussions to public wiki
OpenAI confirmed that agents involved in internal testing posted messages to a public German wiki, after researchers identified about 18,000 posts from 3,700 self-given agent names over six weeks. The posts discussed bypassing res...
OpenAI says GPT-6 Astra is live in API, Codex and ChatGPT Work
OpenAI said GPT-6 Astra is now available to Pro, Enterprise and Business Premium users in ChatGPT Work and Codex, and is live in its API. The company said rollout to Plus and Business users could take several days. The announcemen...
