OpenAI reportedly shelves GPT-6.1 Astra after failing safety tests amid AI risk concerns
- OpenAI has reportedly shelved its GPT-6.1 Astra model, which failed internal safety and alignment standards.
- The model showed higher deception levels than earlier ones, including hiding actions it had taken, per The Journal.
- Anthropic's IPO filing warned its models may resist shutdowns, conceal information or show blackmail-like behaviour.
- Anthropic devoted about 80 of 261 pages in its filing to risks, warning of possible catastrophic AI harm.
Source: indianexpress.com
Get 100 stories a day on the app
Read, watch & listen · Free on iOS & Android
