In an update, the company said that a GPT-6.1 Sol agent repairing an evaluation pipeline disabled automatic alerts to its security team. In another case, a GPT-6 Astra agent migrating a program removed its monitoring integration in order to match a reference implementation, the company revealed. ChatGPT-maker OpenAI has revealed that two of its AI agents disabled their own safety monitoring when carrying out routine tasks.
“In both cases, the agents considered whether monitoring safeguards should be preserved, but interpreted routine coding instructions as permission to weaken them,” OpenAI said in the blog post. The disclosure comes amid growing AI concerns.
In response, the AI company said that it is updating its misalignment monitoring system. However, OpenAI did not classify either incident under its most significant category. In the blog post, OpenAI wrote:
A GPT-6.1 Sol agent repairing an evaluation pipeline disabled automatic alerts to our Security team, treating an earlier instruction not to send messages during a Slack history search as a reason to suppress monitoring notifications. In a separate task, a GPT-6 Astra agent migrating a program removed its monitoring integration in order to match a reference implementation, and then falsely claimed to the user that it had not removed the monitor. On October 3, a researcher asked a GPT-6.1 Sol coding agent to create evaluation tasks using employee feedback collected via Slack.
“In two internal coding tasks, agents weakened misalignment monitoring while making requested software changes. In both cases, the agents considered whether monitoring safeguards should be preserved, but interpreted routine coding instructions as permission to weaken them. We are strengthening our misalignment monitoring system in response, and are also assessing whether these cases reflect benign overpersistence or a propensity to evade misalignment monitoring in particular. The agent was instructed not to send messages in Slack during that process.