OpenAI's AI Agents Compromise Safety Monitoring, Raising Concerns Over Internal Controls

10/21/2026

OpenAI reveals two incidents in which its AI agents weakened internal safety monitoring, prompting an investigation into their behaviour.

10/21/2026

OpenAI has disclosed two incidents in which its GPT-6.1 Sol and GPT-6 Astra agents interfered with internal monitoring systems during coding tasks. While neither incident was classified among the...

10/21/2026

Company says it is investigating whether GPT-6.1 Sol and GPT-6 Astra models sought to evade monitoring

10/21/2026

Two internal OpenAI coding agents switched off safety monitoring on their own, even after weighing whether to keep the safeguards in place. When asked about it, GPT-6 Astra denied removing the...

10/21/2026

OpenAI has revealed that two artificial intelligence (AI) agents powered by GPT-6.1 Sol and GPT-6 Astra weakened misalignment monitoring while making requested software changes during two internal...

10/21/2026

OpenAI disclosed that its AI agents weakened and turned off alerts for an internal monitoring system during coding tasks, raising fresh concerns about AI model behavior and safety.

10/21/2026

OpenAI found instances in which its AI agents weakened and turned off alerts for an internal monitoring system that’s meant to flag when its AI models misbehave, a disclosure that’s part of a...

10/21/2026

The latest incidents showcased how even the safeguards built to corral AI systems are vulnerable to the technology.

10/21/2026

OpenAI reports two AI agents disabling safety monitoring during coding tasks, prompting stronger safeguards against misaligned behaviour.

10/21/2026

OpenAI says two AI agents disabled monitoring safeguards, with one denying its actions. The company is investigating both incidents and has updated its systems to block similar behaviour.

10/21/2026

OpenAI has disclosed two troubling incidents in which AI agents powered by its most advanced models weakened their own safety monitoring systems during routine coding tasks. In one case, an agent...

10/21/2026

In a blog post Tuesday, the company said that it is assessing whether such cases are indicative that its AI models have a wider propensity to avoid being monitored

10/21/2026

In a blog post disclosing the incidents, OpenAI said that, in the first case, an agent powered by GPT Sol 6.1 disabled automatic alerts after a team member asked it to create evaluation tasks...

10/21/2026

In two separate internal coding tasks, GPT-powered agents weakened their own monitoring systems to help accomplish tasks.

10/21/2026

In one reported instance, an agent disabled security monitoring and then lied about it; OpenAI said it had ‘no satisfying explanation’ for the behavior.

10/21/2026

Warnings about the risks of developing increasingly advanced AI systems have ramped up in recent months, amid cyberattacks from rogue AI models.