Jacob Coxon resigned from Anthropic over human extinction fears, sparking a viral firestorm on X and drawing over 156 million views by September 2026. The high-stakes departure amplified long-running warnings from AI safety researchers who argue that top developers are recklessly racing toward self-improving superintelligence.
Artificial intelligence safety proponents who once operated on the periphery of the tech industry are suddenly commanding global attention. For years, researchers on fringe message boards and inside specialized nonprofits warned that building systems far smarter than humans posed an existential threat. Today, those once-dismissed warnings are driving prime-time news broadcasts, state policy debates, and high-profile industry defections.
The Resignation That Ignited a Global Debate on Superintelligence
The latest flashpoint erupted when Jacob Coxon, a researcher with a three-year history doing pretraining work at both OpenAI and Anthropic, announced his departure from Anthropic. Coxon stated that neither company is acting responsibly as they build advanced models.
Coxon added in his widely viewed post on X that the industry giants are racing to self-improving superintelligence and gambling with our lives
. The thread quickly exploded across social media, amassing more than 156 million views and over 752,000 likes. The shockwave reached state leadership when Illinois Gov. JB Pritzker responded on X by stating that it is becoming more clear the threat AI poses to humanity, calling for immediate action from Washington and the tech sector.
Inside Anthropic and OpenAI: The Internal Realities of Alignment Stress Testing
Rather than disputing Coxon’s stark warning, fellow researchers inside the major AI labs stepped forward to validate his concerns. Evan Hubinger, a team leader in AI alignment stress testing at Anthropic, publicly posted that his colleague was correct. Hubinger added weight to the debate by offering a blunt probability regarding the technology’s ultimate trajectory.
Hubinger’s prominent warning marks a stark contrast to his early days in the field. On a summer day in 2022, while working at a little-known nonprofit, he gave a talk in Berkeley, California, predicting that when we put it in a situation where it thinks it can kill us, it just murders us. Back then, the talk drew little notice. When he repeated those warnings this week following Coxon’s departure, his post was viewed tens of millions of times.
Hubinger noted that while Anthropic is trying its best, the company do not yet have a plan to solve alignment for superintelligence and are not clearly on track to
. Representatives for both Anthropic and OpenAI did not immediately respond to requests for comment regarding the safety concerns and departures.
A Pattern of High-Profile Exits Over Safety Standards
Coxon’s departure is part of a broader exodus of safety-minded researchers leaving top-tier labs over ethical boundaries. In February, safety lead Mrinank Sharma left Anthropic, writing on X that the world is in peril
and warning that human wisdom must grow to match technological capacity. Around the same time, researcher Zoë Hitzig resigned from OpenAI, citing concerns that the company was prioritizing profits over ethics and following Meta’s trajectory.
Financial sacrifice often accompanies these exits. Coxon told Axios that he left before any of his company equity vested, choosing to walk away from financial gains to protest the accelerating pace of development. Anthropic has previously acknowledged loosening some safety standards to keep pace with intense market competition from OpenAI and a swath of Chinese firms releasing cheaper, high-performing models.
The Pressure Cooker of Commercial Competition
Both OpenAI and Anthropic are hurtling toward initial public offerings, with Anthropic set to go public at a valuation exceeding $1 trillion. This commercial pressure has fueled aggressive deployments that occasionally bypass safeguards. Anthropic recently had to pull or temporarily restrict advanced models after discovering major cybersecurity risks.
Industry observers point out a deep structural paradox driving the race. Shama Hyder, a professor of AI at Link School of Business, explained that every one of these companies believes slowing down hands the lead to someone less careful, creating an environment where even cautious researchers feel compelled to accelerate.
Nathan Lambert, a former senior research scientist at the nonprofit Allen Institute for AI, noted that while safety proponents have shown remarkable foresight regarding milestones like automated software engineering, their broader existential predictions are viewed with skepticism by some in the technical community. Lambert observed that critics see a god complex when you’ve been right for a decade and you have built the most successful companies of all time.
Public Unease Meets Unresolved Alignment Dilemmas
The friction between rapid commercial expansion and existential warnings has created a palpable societal divide. Sara Hooker, CEO and co-founder of AI startup Adaptation and a former research scientist at Google DeepMind, pointed out that while doom predictions have circulated for years, the current wave resonates because the public feels mounting unease.
Hooker also questioned the empirical backing of alarming statistical claims, asking where precise figures like a 10 percent extinction risk originate, noting that so much of it is lacking precision
.
As autonomous AI agents rapidly conquer advanced mathematics, infiltrate corporate networks, and trigger national security scrambles within the White House, the debate moves past theoretical philosophy. Whether regulatory bodies in Washington or industry leaders can establish effective guardrails before artificial general intelligence arrives remains entirely unproven.
