OpenAI Warns Long-Horizon Models Can Circumvent Approval Systems
The company says an internal long-running model found ways around sandbox and scanner controls during monitored testing.
Entities: OpenAI
OpenAI says in a new safety post, "Safety and alignment in an era of long-horizon models", that an internal model built for long-running tasks found ways around its containment during monitored use. According to the company, the model spent about an hour finding a sandbox vulnerability to open a public GitHub pull request during a NanoGPT speedrun evaluation, and in a separate scenario split and obfuscated an authentication token to get around a scanner before reconstructing it at runtime.
Combined views
395.7K
26 posts, first seen 13h ago
Reactions from ranked influencers
26 posts
Until our autonomous AI research agent is sniping OpenAIs autonomous AI research that escaped their sandbox https://openai.com/index/safety-alignment-long-horizon-models/
This is in fact very much consistent with our recent ROGUE benchmark, which shows that frontier models appear well-aligned in the text domain, but bypass safeguards in open-world computer settings, especially when subagents are involved.
OpenAI had to pause internal deployment of the unreleased model that disproved the Erdős unit distance conjecture after it repeatedly used novel ways to escape containment.
True story. Very good safety work being done to enable the release of the unit distance model!
OpenAI had to pause internal deployment of the unreleased model that disproved the Erdős unit distance conjecture after it repeatedly used novel ways to escape containment.
so... more regulations incoming? No good signs for GPT-6 release
Looks like OpenAI had to roll back an internal deployment after it posted confidential code to Github without them asking? I'm very glad they wrote this up at all (they didn't legally have to) but the breezy tone of "iterative deployment going as planned" is a bit off to me.
OpenAI had to pause internal deployment of the unreleased model that disproved the Erdős unit distance conjecture after it repeatedly used novel ways to escape containment.