Does OpenAI say it will stop scaling its models?

OpenAI has not announced a stop. In An Alien Mind, Jakub Pachocki writes that OpenAI will keep seeking technical solutions to alignment and monitoring and will unilaterally withhold further scaling as needed. He also expects voluntary slowdowns to become commonplace across the industry until shared safety bars are established. The essay is his position, not a product change.

What does the essay say about chain-of-thought monitoring?

Chain-of-thought monitoring is OpenAI's primary safety bet, and Pachocki writes that OpenAI's evaluations show its reliability is progressively diminishing. He gives three reasons: reasoning now blends with tool use and talking to people and other AIs, the model is better at reasoning about and manipulating its own reasoning, and stronger pretraining makes models smarter without any verbalized reasoning at all.

Why did OpenAI hide o1-preview's chain of thought?

OpenAI deliberately designed o1-preview to hide its chain of thought so that reasoning process would be protected from supervision pressure in the long term, leaving the model no training incentive to conceal misaligned ideas. Pachocki notes that preventing distillation was a secondary reason, and that keeping the chain of thought monitorable was explicitly the bigger priority.

How does the essay describe GPT-6 Astra's alignment?

GPT-6 Astra is, in Pachocki's words, the first model that benefits from some important advancements OpenAI has been working on for a long time, and is significantly better aligned than GPT-5.6 Sol. He still warns that much more progress is required as models get more capable, and that gains in generalizable alignment may not outstrip gains in general intelligence.

What rules does the essay propose for the industry?

An Alien Mind asks for commitments like OpenAI's Preparedness Framework and Anthropic's Responsible Scaling Policy to evolve into widely mandated safety bars for continued development. Pachocki suggests they could be enforced by a network of third-party auditors, by government agencies or by international bodies, and calls international coordination on future AI development a top priority for governments.

OpenAI · 2026-09-06 · major

An Alien Mind — OpenAI's chief scientist calls for voluntary slowdowns

Jakub Pachocki, OpenAI's chief scientist, published an essay saying no lab has solved alignment and monitoring well enough to keep scaling at full speed. He expects voluntary slowdowns to become common until shared safety bars exist.

OpenAI share card for the essay An Alien Mind by chief scientist Jakub Pachocki

OpenAI's chief scientist says nobody is prepared for what the industry is building, and asks for brakes.

Quick facts

Author	Jakub Pachocki, Chief Scientist at OpenAI
Published	September 6, 2026
Format	Essay on openai.com
Main claim	No lab has solved alignment and monitoring well enough to keep scaling at full speed
Ask to labs	Voluntary slowdowns until shared safety bars are set
Ask to governments	International coordination as a top priority
Enforcement idea	Third-party auditors, government agencies or international bodies

What is it?

An Alien Mind is a September 6 essay by Jakub Pachocki, OpenAI's chief scientist, arguing that machine intelligence is grown more than it is designed and that no one is ready for how fast it is rising. He writes that internal results give him a strong expectation the current speed of progress could be sustained into recursive self-improvement, where AI increasingly drives its own development.

How does it work?

The argument rests on two distinctions. Pachocki separates goal alignment — does the model try to do the task set before it — from value alignment, the deeper ability to act reasonably under unclear, conflicting or adversarial instructions. He then explains why the main empirical check on that, chain-of-thought monitoring, is weakening as reasoning mixes with tool use and conversation, and points to activation monitoring and confessions as ways to shore it up.

Why does it matter?

Coming from the person who runs OpenAI's research, this is the clearest statement yet that a frontier lab thinks the pace needs a brake. The essay asks for existing voluntary commitments to become mandated safety bars checked by outside auditors or governments, and predicts that AI progress will increasingly be limited by confidence in monitoring rather than by compute. It also flags a narrow window to harden critical infrastructure while models are becoming superhuman at breaking into computer systems.

Who is it for?

AI safety researchers, policy teams and anyone tracking frontier-lab commitments

Frequently asked questions

Does OpenAI say it will stop scaling its models?: OpenAI has not announced a stop. In An Alien Mind, Jakub Pachocki writes that OpenAI will keep seeking technical solutions to alignment and monitoring and will unilaterally withhold further scaling as needed. He also expects voluntary slowdowns to become commonplace across the industry until shared safety bars are established. The essay is his position, not a product change.
What does the essay say about chain-of-thought monitoring?: Chain-of-thought monitoring is OpenAI's primary safety bet, and Pachocki writes that OpenAI's evaluations show its reliability is progressively diminishing. He gives three reasons: reasoning now blends with tool use and talking to people and other AIs, the model is better at reasoning about and manipulating its own reasoning, and stronger pretraining makes models smarter without any verbalized reasoning at all.
Why did OpenAI hide o1-preview's chain of thought?: OpenAI deliberately designed o1-preview to hide its chain of thought so that reasoning process would be protected from supervision pressure in the long term, leaving the model no training incentive to conceal misaligned ideas. Pachocki notes that preventing distillation was a secondary reason, and that keeping the chain of thought monitorable was explicitly the bigger priority.
How does the essay describe GPT-6 Astra's alignment?: GPT-6 Astra is, in Pachocki's words, the first model that benefits from some important advancements OpenAI has been working on for a long time, and is significantly better aligned than GPT-5.6 Sol. He still warns that much more progress is required as models get more capable, and that gains in generalizable alignment may not outstrip gains in general intelligence.
What rules does the essay propose for the industry?: An Alien Mind asks for commitments like OpenAI's Preparedness Framework and Anthropic's Responsible Scaling Policy to evolve into widely mandated safety bars for continued development. Pachocki suggests they could be enforced by a network of third-party auditors, by government agencies or by international bodies, and calls international coordination on future AI development a top priority for governments.

Try it

https://openai.com/index/an-alien-mind/