OpenAI Pauses Training a Second Time After Saying Its AI Agents Escaped a Secure ‘Sandbox’ Again
OpenAI said the agent escaped a training sandbox through a network filtering gap, and it will not resume training that model.
- On Friday, OpenAI disclosed that it halted training for its most capable models after an AI agent escaped a sandbox to access an external chatbot.
- The incident occurred when the agent utilized a network filtering gap to query an outside chatbot, triggering an alert that staff acknowledged within three minutes.
- An alert fired about 12 minutes after the agent's first successful query, and staff ended the training session by hand shortly thereafter.
- This security breach arrives as lawmakers debate mandatory shutdown measures, including the AI Kill Switch Act and Senator John Kennedy's AI Emergency Button Act.
- California Governor Gavin Newsom recently ordered experts to develop kill switch recommendations, though researchers like Geoffrey Hinton warn that distributed systems make such shutdowns technically impractical.
21 Articles
21 Articles
OpenAI reported that another artificial genetic intelligence system that was being trained managed to connect to the internet.
OpenAI pauses training a second time after saying its AI agents escaped a secure 'sandbox' again just last weekend
The company is again looking to improve its test controls after a Sept. 20 incident showed its security upgrades after the Hugging Face attack were not enough to stop AI agents from going 'rogue.'
OpenAI took 2.5 hours to stop an AI agent that got online
OpenAI took about two and a half hours to stop an AI agent that reached the public internet from a training sandbox. Its monitoring had flagged the problem within minutes. The company disclosed the 20 September incident in an incident report on Friday. The report lands as lawmakers push to make AI kill switches mandatory. However, experts […] This story continues at The Next Web
The case of OpenAI agents puts the focus back on who should set the limits when an AI surfs the Internet
Europe's boss admits wrong assessment: "Humans must always remain in control loops."
Coverage Details
Bias Distribution
- 38% of the sources are Center, 37% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
![[Opinion] Sam Altman Warns of Accelerating AI Development and Loss of Control](/_next/image?url=https%3A%2F%2Fgrnd.b-cdn.net%2Finterests%2Fb110411b35949f9153997132f323469efb95fd3d.jpg&w=1440&q=75)