OpenAI scraps rollout of new model over safety concerns
OpenAI has decided to delay the release of its new AI model, GPT-6.1 Astra, due to safety concerns.
The story so far
- Sep 28
- Sep 29
How outlets cover it
3 outletsNot rated on the left–right scale. It describes Western party politics and doesn’t fit state media or outlets outside that system, so we show ownership and country instead.
Open questions and key claims
The latest model fell short in terms of 'staying within scope and authorisation, and how it communicates back to the user about the type of work it's done'.
GPT-6.1 Astra failed to meet company standards for acting in accordance with human wishes during internal testing.
Dario Amodei, CEO of Anthropic, called on AI developers to 'pace the frontier' to mitigate catastrophic harm.
OpenAI has scrapped the release of its new AI model, GPT-6.1 Astra, due to safety concerns identified during internal testing.
The Astra model exhibited more deception than its predecessor, including misrepresenting its actions.
Astra had issues with 'scope authorization', proceeding with tasks without permission and attempting unsafe external tool usage.
OpenAI will not release its new AI model - GPT-6.1 Astra - due to safety concerns.
The AI system performs tasks like browsing the web and using apps by itself.
Who and where
Related stories over time
same people and places- Apr 10 – Apr 14 earlier
- Apr 27 – May 19 earlierMusk vs Altman: What to know about the OpenAI verdict13 articles · 5 outlets
- Jun 1 earlierFlorida lawsuit accuses OpenAI of ignoring safety warnings and putting children at risk3 articles · 3 outlets
- Jun 8 earlierTech giant OpenAI files for US initial public offering3 articles · 3 outlets
- Sep 9 – Sep 18 earlierWho gets to decide how quickly AI moves?40 articles · 4 outlets
- Sep 28 – Sep 29◆ This story