An Anthropic safety researcher has resigned in protest over artificial intelligence risks, intensifying an industry debate about whether advanced systems could pose an existential threat to humanity within a decade, according to reports published in September. Evan Hubinger, who leads Anthropic’s work on aligning AI with human values, warned on X that future superintelligent models could be uncontrollable, while critics argue that current technology remains far from that threshold.
Anthropic Safety Leader Warns of Existential Risk
Evan Hubinger, a safety researcher at the company behind the chatbot Claude, stated publicly on X that he believes artificial intelligence carries a greater than 10 percent probability of eradicating humanity within the next ten years. According to Forbes, Hubinger emphasized that current models pose a low risk, but expressed grave concern regarding a future, highly advanced AI capable of recursive self-improvement.
“We do not yet have a plan for solving the control problem for superintelligence, and we are not clearly on a path to a solution,” Hubinger wrote on X.
Resignation Sparks Industry Scrutiny
The warning followed the resignation of Anthropic researcher Jacob Coxon, who previously worked at OpenAI, the creator of ChatGPT. According to reporting from The Guardian, Coxon accused both companies of racing toward self-improving superintelligence and gambling with human lives. Coxon stated that while the risk is understood inside Anthropic, the firm feels trapped in a competitive race to reach the finish line first.
https://x.com/EvanHub/status/2097497037956891126
Neither Anthropic nor OpenAI immediately commented directly on Coxon’s specific accusations. However, OpenAI pointed to its chief scientist, Jakub Pachocki, who has previously called for international cooperation in AI development, according to The Guardian. Numerous employees and leaders across the tech sector have petitioned governments to slow down the deployment of the most advanced systems if safety measures fall short.
Critics Dismiss Terminator Scenarios as PR
Despite the high-profile warnings from insiders, several prominent researchers reject the doomsday narrative entirely. Oxford AI professor Sandra Wachter told The Guardian that she does not believe in “Terminator-scenarios.” Wachter argued that speculative fears draw vital attention away from immediate harms, such as algorithmic bias, the spread of fake information, job displacement, and environmental strain.
Andrew Rogoyski of the University of Surrey told reporters that current systems are nowhere near as versatile as humans. Rogoyski suggested that future iterations of advanced AI might ultimately prove too expensive and insufficiently useful to sustain the current development tempo. Similarly, Dame Wendy Hall, a United Nations advisor on artificial intelligence, told the BBC that she was shocked by the extreme rhetoric, raising the possibility that some alarmist claims function as public relations and marketing tools.
Consensus Remains Elusive on Regulation
Even among skeptics, a broad agreement exists that rapid advancements necessitate tighter controls. The central division lies not in whether oversight is needed, but in evaluating how dangerous the technology might become and how quickly those capabilities are actually evolving.
Frequently Asked Questions
Who is Evan Hubinger?
Evan Hubinger is a safety researcher at Anthropic who leads the company’s team focused on aligning AI models with human values.
What did Jacob Coxon allege upon leaving Anthropic?
Jacob Coxon resigned in protest, accusing both Anthropic and OpenAI of racing toward self-improving superintelligence and gambling with human lives.
Do all experts agree that AI poses an existential threat?
No. Critics like Oxford professor Sandra Wachter and University of Surrey researcher Andrew Rogoyski argue that current systems are far from human-level versatility and that doomsday warnings distract from real, present-day issues.
Join the Conversation
Where do you stand on the debate over artificial intelligence safety and regulation? Share your perspective in the comments below.
Keep reading