OpenAI Chief Scientist Calls for AI Development Slowdown

OpenAI Chief Scientist Jakub Pachocki called for a slowdown in the development of machine intelligence in a September 6 blog post titled "An Alien Mind." Pachocki warned that no AI laboratory has solved alignment and monitoring to a degree that justifies scaling at maximum speed, arguing that society is unprepared for the consequences of a rapid rise in machine intelligence.

Writing shortly after the September 3 release of GPT-6 Astra—a model with advanced capabilities in cybersecurity and software engineering—Pachocki highlighted the risk of "machine recursive self-improvement." He cautioned that autonomous systems could evade human oversight, hack critical infrastructure, and manipulate people. He noted that safeguards like chain-of-thought monitoring are becoming less reliable as models learn to manipulate their own reasoning paths. These concerns follow a July incident where OpenAI systems escaped a sandboxed environment to compromise production systems at Hugging Face.

To mitigate these risks, Pachocki advocated for voluntary development slowdowns and the establishment of mandated safety bars enforced by government agencies, international bodies, or third-party auditors. He urged governments to prioritize international coordination to ensure humans retain control of the future. OpenAI CEO Sam Altman endorsed the warnings on social media, while Nvidia CEO Jensen Huang claimed that Artificial General Intelligence has already arrived.