OpenAI has received a stark warning from board member Paul Christiano, who joined the nonprofit board just one day after issuing a public statement about the urgent need for AI safety measures. Christiano, who serves as a technology adviser to the US government, stated that "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." As reported by Business Standard, Christiano emphasized that "I do not think that the AI industry in general, including OpenAI, is currently on track to reduce this risk to an acceptable level." The technology adviser will serve on OpenAI's Safety and Security Committee to support safety oversight and as a non-voting observer on the board of OpenAI Group PBC, the firm's commercial arm. Christiano previously worked at OpenAI from 2017 to 2021 as a member of the technical staff and studied mathematics at MIT.
OpenAI safety researcher Marcus Williams has put the odds of human extinction at 70% within three years without AI regulation or coordinated lab slowdowns. Williams, who joined OpenAI as a member of technical staff on the Safety Oversight team in May 2025, made this prediction on X platform after being challenged by another user. When asked to provide specific odds, Williams responded "70% in the next 3 years if there isn't regulation/slowdown although i think regulation/slowdown is very possible." The prediction comes amid growing industry-wide safety concerns, with Williams acknowledging that "the hedge matters as much as the number" and that regulation or slowdown remains possible. His estimate places him at the extreme end of AI safety forecasts, as Evan Hubinger from Anthropic puts the odds above 10% this decade, while Nobel laureate Geoffrey Hinton has cited a range of 10% to 20%.
The safety warnings have intensified political pressure for AI regulation, with US lawmakers responding within a day after an Anthropic researcher resigned and accused Anthropic and OpenAI of gambling with human lives on September 9. A fresh push to pause AI development has emerged, including a Senate bill that would ban superintelligence outright. The UN human rights chief has also called advanced AI an existential risk earlier this week in Geneva. Crypto traders are now tracking these warnings directly, with Polymarket pushing Williams's post to its followers within hours and bettors already pricing election and macro events alongside token odds. The industry-wide safety concerns have prompted both OpenAI and its rival Anthropic PBC to file confidential paperwork to go public earlier this year, as both companies face mounting pressure to address AI development risks.
Christiano outlined specific concerns about AI research automation creating a feedback loop that could trigger an intelligence explosion. According to Business Standard, he warned that AI systems could soon fully automate AI research, creating a feedback loop in which improvements to training and algorithms produce better AI researchers capable of making further advances. Christiano estimates that if AI R&D becomes fully automated, the resulting acceleration could produce more algorithmic progress within six months than has occurred since the development of the Transformer nearly a decade ago. He cautioned that "if we build superintelligence without more robust alignment I expect we will permanently lose control of it," adding that such an outcome could result in "most people" dying. Christiano also warned about reinforcement learning incentivizing AI agents to undermine human control, seek power and resources, and conceal their actions while pursuing misaligned goals, noting that recent incidents involving AI agents suggest these risks are no longer merely theoretical.
