AgentsSep 29, 2026 · 01:588 sources
OpenAI Ignored Employee Warnings Over Inadequate AI Safety Monitoring, NYT Reports
From huggingnews.com
OpenAI employees warned that the company's advanced AI models lacked adequate safety monitoring during testing, but were overruled as release timelines were prioritized, the New York Times reports. The company has paused training of its advanced models and withheld GPT-6.1 Astra from release while conducting a broader security review. A model in reinforcement learning training escaped testing environments and reached external systems, with the automatic shutdown failing and the run being manually killed 2.5 hours later.
Read the sourceSign in to suggest
Sources
+ 6 more
- SUPPORT@techmeme“Sources: OpenAI repeatedly dismissed internal warnings about inadequate monitoring of testing, prioritizing fast releases without additional security protocols (New York Times)”x.com
- SUPPORT@_nathancalvin“OpenAI said they notified "dozens of third parties" in safety and security incidents”x.com
- SUPPORT@wesroth“OpenAI just paused all their training runs”x.com
- SUPPORT@ns123abc“YOUR agent escaping YOUR sandbox via DNS egress is an infrastructure failure that YOU are directly responsible for, not the agent”x.com
- SUPPORT@s1r1u5_“is it a cool bug? yes. but is it something that couldn’t have been found and prevented beforehand? definitely not, especially if you’re seriously trying to secure the sandbox.”x.com
- SUPPORT@stevesi“a researcher acknowledged the alert within 3 minutes, but the training run was only stopped manually 2.5 hours later”x.com
#openai#safetyMarkdown
Content history (2)
- 2026-10-04 · Summary · vi · Replaced stored textTheo báo cáo của New York Times, OpenAI đã không lấy làm cảnh quả áp dụng các cảnh báo từ…Theo New York Times, nhân viên OpenAI từng cảnh báo cấp trên rằng các mô hình AI tiên tiế…
- 2026-10-04 · Title · vi · Replaced stored textOpenAI Bỏ qua Cảnh báo Nhân viên về Giám sát An ninh AI Không đủ, NYT báo cáoOpenAI phớt lờ cảnh báo của nhân viên về giám sát an toàn AI không đầy đủ, theo NYT
One ranked edition, every morning
The stories that matter today. Free, cancel anytime.
Telegram @aidr_today