OpenAI Chief Scientist Jakub Pachocki has warned that AI capabilities are advancing faster than the safeguards needed to control them.
He said no AI lab has solved alignment and monitoring well enough to continue scaling at maximum speed responsibly.
Pachocki called for voluntary slowdowns and greater coordination on common AI safety standards.
OpenAI Scientist Warns AI Could Outpace Human Control, Calls For Slowdowns
Jakub Pachocki says AI development needs stronger safeguards and shared safety standards to ensure humans remain in control of increasingly capable systems
Summary
Advertisement
OpenAI Chief Scientist Jakub Pachocki has warned that the rapid development of artificial intelligence (AI) could outpace humanity’s ability to keep increasingly capable machines under control.
In a blog post titled “An Alien Mind”, published on September 6, Pachocki said the focus should be on ensuring that the rise of more powerful AI systems does not come at the cost of human control.
Pachocki wrote, “Currently I believe that no lab has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer.”
Pachocki Calls For AI Slowdowns
Pachocki said he expects and hopes that voluntary slowdowns in AI development will become more common until shared safety standards are established.
His concerns go beyond the impact of AI on jobs and productivity. He described a future in which advanced AI systems could operate computers, work with humans and other AI systems, and carry out research with increasing levels of independence.
Advertisement
He said the key challenge is not only to build more capable machines but also to ensure that humans can continue to direct and control them as their abilities grow.
Growing AI Capabilities Raise Safety Concerns
Pachocki also warned about the possibility of an “alien intellect” that could eventually exceed human capabilities. He said unchecked technological progress could risk leaving people behind.
AI alignment broadly refers to making sure AI systems behave in line with human intentions and values. Monitoring becomes increasingly important as models become capable of carrying out longer and more complex tasks.
His comments come as AI companies continue to develop increasingly capable frontier models. OpenAI recently introduced GPT-6 Astra, highlighting advances in areas including computer use, browsing, software engineering, cybersecurity, science and professional work.
OpenAI’s safety assessment for Astra said the model represents a significant increase in cybersecurity capabilities, while the company has also strengthened safeguards and monitoring of model behaviour.
Advertisement
Pachocki argued that AI laboratories alone cannot address the risks. He called for stronger cooperation between governments and AI companies, as well as common safety standards, to ensure that humans “remain in control of the future and are not left behind by unchecked progress.”