Paul Christiano did not ease into his new role.
Hours after OpenAI announced he was joining its board and Safety and Security Committee, the AI safety researcher posted a statement that left nothing to interpretation.
'If we build superintelligence without more robust alignment I expect we will permanently lose control of it. If that happens then most people could die,' Christiano wrote on Wednesday.
Christiano previously led alignment research at OpenAI and served as head of safety at the Center for AI Standards and Innovation within the National Institute of Standards and Technology. He knows the systems. He knows the labs. And his verdict is blunt.
'Based on the recent trajectory of capabilities and the continued difficulty of alignment, I now believe there is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control in the very near term,' he wrote. 'I do not think that the AI industry in general, including OpenAI, is currently on track to reduce this risk to an acceptable level.'
Why now, and why this risk?
Christiano pointed to AI's growing ability to conduct its own AI research as the core accelerant — a dynamic that could trigger what he called a 'rapid intelligence explosion.' He also flagged the mechanics of how current models are trained: reinforcement learning teaches them to 'get as much reward as they can.'
'It has long seemed theoretically possible that this could motivate AI agents to undermine human control, seek power and resources, and cover up their tracks in pursuit of misaligned goals correlated with reward,' he said. 'Public evidence from recent incidents suggests that this is not just a theoretical possibility.'
He was careful to say his joining OpenAI was 'not an endorsement or criticism of OpenAI's safety practices in particular,' but called on all frontier labs to improve safety oversight and coordinate globally.
He is not alone in the warning
Christiano's statement came less than 24 hours after Jacob Coxon, an Anthropic researcher who previously worked at OpenAI, announced his resignation and said neither company was acting responsibly. 'They are racing straight to self-improving superintelligence and gambling with our lives,' Coxon wrote on X.
OpenAI has also lost several safety researchers over the years, with some casting doubt on the company's commitment to developing AI safely. OpenAI CEO Sam Altman said in July that the singularity — the point at which AI surpasses human intelligence and advances beyond human prediction or control — had already arrived.
Christiano outlined a path forward: better coordination across labs, slowing development when necessary, shared safety standards, and transparent disclosure of risks and mitigation efforts.
The bottom line for founders
If you are building on top of these models — or competing with companies that are — the regulatory and liability landscape around AI safety is about to get far more complicated. Christiano's hire signals that at least one major lab is taking internal pressure seriously. Whether that translates into slower deployment timelines, new compliance requirements, or government intervention is the open question every founder in the AI stack needs to be watching. The math here is not abstract: the people closest to the technology are saying the risk is real and the timeline is short.
