OpenAI Safety Researcher Puts Human Extinction at 70% Within 3 Years
A member of technical workers on OpenAI’s Safety Oversight group put the percentages of human extinction at 70% inside three years until AI labs decelerate or regulators step in.
Marcus Williams, who joined the corporate in May 2025, made the decision on X. He posted the warning first, then connected a quantity when one other person pushed him for one.
How the Human Extinction Estimate Landed
Without AI regulation or a coordinated slowdown between labs, Williams wrote, human extinction within the subsequent few years appears very doubtless.
Another person then turned it right into a wager, providing $10,000 in Microsoft shares towards OpenAI inventory over 5 years. Williams answered with a proportion.
“70% within the subsequent 3 years if there isn’t regulation/slowdown though i believe regulation/slowdown may be very attainable”
However, the hedge issues as a lot because the quantity. Williams calls a regulation or a slowdown very attainable, so he doesn’t deal with the result as settled.
Washington Is Already Moving on AI Risk
The submit lands in every week that pushed lab security into mainstream politics. On September 9, an Anthropic researcher resigned and accused Anthropic and OpenAI of playing with human lives.
US lawmakers responded inside a day. A contemporary push to pause AI adopted, together with a Senate invoice that will ban superintelligence outright.
Voices outdoors the labs have echoed the alarm. The UN human rights chief referred to as superior AI an existential danger earlier this week in Geneva.
Williams nonetheless sits at the acute finish of those forecasts. Evan Hubinger, who leads alignment science at Anthropic, places the percentages above 10% this decade. Nobel laureate Geoffrey Hinton has cited a variety of 10% to twenty%.
Meanwhile, crypto merchants now monitor these warnings immediately. Polymarket pushed Williams’s submit to its personal followers inside hours, and its (*3*) and macro occasions alongside token odds.
Therefore, the near-term take a look at is political somewhat than technical. Whether regulators act inside Williams’s three-year window stays open, and his personal reply to this point is that they nonetheless can.
The submit OpenAI Safety Researcher Puts Human Extinction at 70% Within 3 Years appeared first on BeInCrypto.
