OpenAI Safety Researcher Sets 70% Extinction Odds, Keeping Worldcoin (WLD) in Focus

A 70% Call From Inside OpenAI

A stark probability estimate from inside OpenAI is pulling Worldcoin (WLD) back to the center of the AI-risk conversation. Marcus Williams, a member of technical staff on the company's Safety Oversight team, argued on X on September 10, 2026 that, without AI regulation or a coordinated slowdown among frontier laboratories, human extinction within the next few years “seems very likely.” Williams, who joined OpenAI in May 2025, initially offered no figure; the number surfaced only after another user pressed him and proposed a wager of $10,000 in Microsoft shares against OpenAI stock over five years. His reply set the odds at 70% within three years absent regulation or a slowdown, while adding that he considers either intervention “very possible.” That hedge matters as much as the headline number: Williams explicitly frames the outcome as conditional on policy inertia rather than predetermined. He does not treat the result as settled, only as likely under a no-action scenario. The full exchange is preserved in his original September 10 warning and in the reply fixing the 70% figure. For the token tied to Sam Altman's identity venture, the relevance is structural rather than philosophical. Worldcoin, the blockchain-based proof-of-personhood network, has repeatedly traded as a sentiment proxy for OpenAI news flow because it represents the most direct crypto exposure to Altman's AI franchise, and our consolidated coverage of the project sits on the Worldcoin topic hub. Within hours of the post, prediction-market accounts amplified the warning to their own follower bases, and traders began weighing what a genuine regulatory clampdown on frontier laboratories would mean for AI-adjacent digital assets. Our desk's reading of prior episodes is consistent: OpenAI headlines on governance, funding or pacing have produced sharper short-term repricing in WLD than most project-specific announcements, making this 70% call a live market input rather than an abstract debate.

his original September 10 warninghttps://x.com/Marcus_J_W/status/2098078076299366684?ref_src=twsrc%5Etfw

Washington Response Raises the Stakes

The warning landed in a week when laboratory safety moved from research blogs into mainstream politics. On September 9, an Anthropic researcher resigned publicly and accused both Anthropic and OpenAI of gambling with human lives. US lawmakers responded within a day: a fresh push to pause frontier development gathered pace, including a Senate bill that would ban superintelligence outright. Outside the labs, the United Nations human rights chief called advanced AI an existential risk earlier in the week, speaking in Geneva. Against that backdrop, Williams's figure sits at the extreme end of expert forecasts rather than outside them. Evan Hubinger, who leads alignment science at Anthropic, places the odds above 10% this decade, while Nobel laureate Geoffrey Hinton has cited a range of 10% to 20%. The dispersion among these estimates is itself a datapoint: even the low end now carries double-digit stakes this decade. The market plumbing for these debates already exists: Polymarket, the prediction-market dapp, pushed Williams's post to its own followers within hours, and its bettors routinely price elections and macro events alongside token odds. The same convergence of AI exposure and crypto speculation surfaced when OKX listed 10x pre-IPO perps in Europe on OpenAI and Anthropic. The week's earlier OpenAI datapoint, the reported investment-grade credit ratings pursuit by OpenAI and Anthropic, had already kept the franchise in front of credit and crypto desks alike. For WLD, the cumulative effect is a tightening link between AI-policy headlines and token sentiment. Our earlier reporting tracked Worldcoin slipping 3.2% on Altman's slowdown signal; the current episode stress-tests whether renewed safety alarms cut the other way for a token whose best-known co-founder runs the laboratory at the center of the debate. Traders gauging broader positioning can follow our altcoin market coverage for signs that AI-adjacent tokens decouple from the wider market. Readers tracking the market in real time can follow live spot and futures prices on Bybit.

Worldcoin (WLD) as the AI-Risk Proxy

The load-bearing primary record this week is Williams's own X thread — the warning post and the numbered reply — and COINOTAG's view is that it belongs on the policy calendar, not the price chart. The arc runs one way: lab insiders, legislators and UN officials are converging on the same risk language, and the open question is whether regulators act inside Williams's three-year window. Until that resolves, sentiment gauges like the fear and greed index will keep translating every OpenAI signal into WLD order flow. Whether that flow turns WLD into a hedge, a casualty or both remains a function of Washington, not of model benchmarks.