Ars Technica · AI· Kyle Orland·· 2026-09-29
OpenAI says planned GPT-6.1 is too insecure to release
OpenAI says planned GPT-6.1 is too insecure to release
AI summary
Similar performance, security trade-offs also seen in current public models.
Selection record
Threshold 76Media and individualsFirst 82Second 87
AdmittedSum of both 169 ≥ twice the threshold 152
- Source tier
- Media and individuals; this tier's threshold is 76
- Pre-filter
- passed:OpenAI取消GPT-6.1发布,涉模型安全
- Why it was chosen
- Cancelling GPT-6.1 over alignment and tool-use safety regressions illustrates concrete trade-offs between performance and safety.
- Same event
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Ars Technica · AI · arstechnica.com