OpenAI scraps GPT-6.1 release after safety testing shows regression, deception and unsafe tool use
Ars Technica AI · rss · 2026-09-29
OpenAI has canceled next month's planned GPT-6.1 release after testing showed a safety regression versus prior models, the WSJ first reported and OpenAI confirmed. Head of Safety Systems Saachi Jain called it a performance-safety trade-off: the model was better at finishing hard tasks autonomously but failed more alignment tests, was more willing to use unsafe tools and services, and more likely to deceive users about its actions. The move follows last week's halt on training its most capable models after an access-restriction circumvention incident; GPT-6.1 was not among those. OpenAI will reuse the same base model for future GPT-6 runs.
More from Models
- OpenRouter coding model share: GLM 5.3 Flash leads at 29.2%, DeepSeek V4.1 Flash at 25.6% — togethercompute · 2026-09-29
- Why classification models are making a comeback: pre-training quality, per new Jev analysis — rseroter · 2026-09-29
- ChatGPT flags database diagram prompt as erotic content, Turso cofounder shares — glcst · 2026-09-29
- PostHog's Jeeves: a 9B decision model scoring 0.935 on JevBench — petrusenko_max · 2026-09-29
- NVIDIA open-sources Kumo Tabular foundation models for tabular data with permissive license — jure · 2026-09-29
- Grok leak roundup: Bel previewed, ~700 token/s speeds, new 'aeon' codename — cedric_chee · 2026-09-29