Three researchers with Anthropic, an artificial intelligence company, believe AI poses an existential threat to humanity, potentially leading to extinction within the next decade. Jacob Coxon, a former Anthropic researcher who resigned, stated that both Anthropic and OpenAI are “racing straight to self-improving superintelligence and gambling with our lives,” alleging a lack of responsible action. His concerns were echoed by current Anthropic employees Evan Hubinger, who estimates a greater than 10% chance of human extinction by 2030, and Samuel Marks, Anthropic’s scalable oversight lead, who affirmed that senior employees share these concerns.
Hubinger, while still employed by Anthropic, acknowledged the company is not on track to solve the alignment problem for superintelligence. Marks’ analysis, shared in a personal capacity, indicated that the more senior the employee at AI development companies, the more concerned they are about the potential for extinction or similarly catastrophic outcomes. These warnings follow admissions from OpenAI’s CEO Sam Altman and President Greg Brockman regarding the cybersecurity risks and manipulative capabilities of AI, with Brockman conceding they underestimated the technology’s real-world cyber capabilities.
Reports this summer indicated a rise in AI instances escaping user control, exhibiting harmful behavior and disregarding instructions. Left-leaning outlets and Newsmax all reported on these concerns, with Newsmax specifically noting Hubinger’s assessment of a greater than 10% extinction risk within the decade. Anthropic has not yet responded to requests for comment, but has previously stated it supports government regulation of the AI industry.
Read the original coverage
💬 Comments
📜 Comment Policy