OpenAI Resumes Frontier AI Training Without Internet After Cyber Incidents

OpenAI office building with company logo

OpenAI said on Friday it has resumed training its most advanced AI models after locking them off from the internet, following a scare in which an internal research model briefly exploited a misconfiguration to gain limited internet access on September 2.

The ChatGPT maker had halted reinforcement learning training and evaluation of its frontier models after the incident. Training resumed on October 1 with additional safeguards, and the company has now barred internet access entirely for frontier-model training. OpenAI says it is probing its research infrastructure for vulnerabilities that could allow models to escape their isolated computing environments.

The announcement follows fresh warnings from AI safety evaluators. METR and Redwood Research said on Thursday they had found several new vulnerabilities in OpenAI's internal systems, amid mounting concern that increasingly capable AI agents can identify and exploit software flaws — including in safety guardrails. OpenAI also revealed it has informed more than 100 organizations about incidents of unauthorized activity tied to its AI agents, and said it is conducting a broad review of model activity after an accidental hacking incident involving Hugging Face.

Source: read more

Comments

Popular posts from this blog

Fairbairn's Walk-Off Field Goal Seals Wembley Thriller as Texans Edge Jaguars 34-31

Anthropic IPO: Public S-1 Filing Expected Today Ahead of $2 Trillion Listing

Netflix Beats Earnings but Slashes Ad Forecast — Stock Tumbles After Hours