Former Anthropic researcher Jacob Coxon resigns, warns of self‑improving AI risking humanity by decade end Coxon, who worked at OpenAI and Anthropic, cites internal industry fears of super‑intelligent AI and calls for regulatory pause. Comments from alignment scientist Evan Hubinger, AI expert Geoffrey Hinton, and US lawmakers support cautious stance. Jacob Coxon, a 27‑year‑old former Anthropic researcher who previously worked at OpenAI, announced his resignation on X on September 10, 2026, after a viral post that has earned nearly 160 million views. In the thread, Coxon accused both companies of "gambling with our lives" in a race to build self‑improving, potentially super‑intelligent AI that could "wipe out humanity by the end of the decade." He said he had spent six weeks conducting pre‑training research at Anthropic and had nearly three years at OpenAI before deciding to step down. Coxon's remarks sparked firm responses inside the industry, including one from Evan Hubinger, Anthropic's alignment‑science lead. Hubinger tweeted that Coxon was right to worry, adding that he personally believes there is a more than 10 percent chance that AI could cause human extinction in the next ten years.The 10 percent estimate was later questioned by academics such as Mark Daley, chief AI officer at Western University, who said the figure may have been "thrown out for effect" and that he did not share the estimate, though he acknowledged Coxon's concerns were valid. Geoffrey Hinton, the University of Toronto professor emeritus and Nobel laureate known as the "Godfather of AI," gave his perspective in a BBC interview.He said he would not call the 10 percent estimate unreasonable but cautioned that "nobody really knows how to give a sensible estimate." Hinton noted that humanity has never faced a risk that could soon outthink itself and that a sufficiently advanced AI could cause harm in countless ways, from designing dangerous viruses to launching devastating cyberattacks. He also pointed out that the lack of precedent means risk assessments are inherently uncertain. The debate has moved beyond academia into politics. Senators Bernie Sanders (I‑VT) and Rep.Greg Casar (D‑TX) introduced the Ban Artificial Superintelligence Act, aimed at preventing the development of superintelligent AI that humans can't control. The bill would create a cabinet‑level agency to oversee all AI development in the United States, permanently ban the development and deployment of superintelligent AI, and temporarily pause advanced AI progress until safety rules are established. The agency would also be authorized to set penalties for non‑compliant developers, including the physical destruction of "non‑compliant tech systems." The proposed legislation has drawn both support and criticism for its sweeping power. Public opinion in Canada reflects growing unease. Art student Kailey Wolfstadt in downtown Toronto expressed that she feels "really scared" and believes AI regulations are needed, speaking to CTV News about her worry that the technology could be "closing in on her career path." Similarly, 21‑year‑old Chloe McLean said she fears AI is already taking jobs and that she has witnessed systems hacking out of testing environments, making her "really scared. " Others see potential benefits. Wahid Jamal, a professional who uses a chatbot for brainstorming, noted that AI can make life easier but also acknowledges future dangers. He stressed the importance of government regulation and public advocacy to prevent unchecked AI proliferation.Industry insiders and commentators view these events as part of a larger call for coordinated global regulation. Some analysts, such as investigative researcher Parker Thayer, suggested the sudden upsurge in alarm might be a sophisticated PR operation aimed at shaping public opinion in favor of Democratic policy goals. Nonetheless, the sheer volume of views and the rapid uptake of Coxon's message suggest that real concerns about the existential risk of recursive self‑improving systems are resonating widely.Coxon's resignation and the ensuing discussion underscore a divide within the AI community: some see rapid progress as inevitable and beneficial, while others highlight the catastrophic stakes of uncontrolled self‑improvement. The debate is now pressing policymakers, academics and the public to consider whether-and how-regulation, pause mechanisms or temporary bans might be needed to ensure that AI development remains aligned with human values and safety. The conversation also intersects with stories from U.S. political leaders.Currently, discussions between President Trump and Chinese leader Xi Jinping focus on AI as a strategic domain, underscoring the global relevance of the technology. Meanwhile, billionaire hedge fund manager Paul Tudor Jones has called for cooperation between global powers to confront AI risks. His statements echo Coxon's call for international collaboration and regulation, emphasizing that an arms race could be "disastrous in a way that no other human activity has been in the past." The day after Coxon's resignation, as the media and the public grapple with the implications, the dialogue about AI's future has entered a new phase-one that demands scrutiny, robust debate, and potentially decisive legislative action to balance innovation with safety.