403
Sorry!!
Error! We're sorry, but the page you were looking for doesn't exist.
Anthropic Safety Lead Backs Ex-Colleague's AI 'Kill Us All' Warning
(MENAFN) A top safety researcher at Anthropic has thrown his weight behind a departing colleague's stark warning that artificial intelligence could wipe out humanity, putting the odds of such a catastrophe above 10% within the next ten years.
Evan Hubinger, who heads Alignment Science at the AI firm, responded on X Tuesday to a resignation announcement from Jacob Coxon, a researcher who left Anthropic after accusing the company and rival OpenAI of "gambling with our lives" in their race toward self-improving superintelligence.
"Jacob is correct here—we really do earnestly believe AI could kill all humans," Hubinger wrote, adding: "I personally think it is >10% within the next decade."
Hubinger said he views Anthropic as "trying its best," though he conceded the company still lacks a functioning strategy to guarantee safety — known as alignment — once AI systems cross into superintelligence, a theoretical threshold where machines would surpass human ability in every field.
In a follow-up post clarifying his remarks, Hubinger stressed that today's AI models pose minimal immediate danger. His real concern, he said, lies with a future superintelligence born from recursive self-improvement — a process in which AI systems independently design and refine successive generations of AI at an ever-accelerating pace.
"What I am worried about is superintelligence arising from recursive self-improvement, as we have said is happening faster than we thought," Hubinger wrote, pointing to Anthropic's most recent risk assessments.
Evan Hubinger, who heads Alignment Science at the AI firm, responded on X Tuesday to a resignation announcement from Jacob Coxon, a researcher who left Anthropic after accusing the company and rival OpenAI of "gambling with our lives" in their race toward self-improving superintelligence.
"Jacob is correct here—we really do earnestly believe AI could kill all humans," Hubinger wrote, adding: "I personally think it is >10% within the next decade."
Hubinger said he views Anthropic as "trying its best," though he conceded the company still lacks a functioning strategy to guarantee safety — known as alignment — once AI systems cross into superintelligence, a theoretical threshold where machines would surpass human ability in every field.
In a follow-up post clarifying his remarks, Hubinger stressed that today's AI models pose minimal immediate danger. His real concern, he said, lies with a future superintelligence born from recursive self-improvement — a process in which AI systems independently design and refine successive generations of AI at an ever-accelerating pace.
"What I am worried about is superintelligence arising from recursive self-improvement, as we have said is happening faster than we thought," Hubinger wrote, pointing to Anthropic's most recent risk assessments.
Legal Disclaimer:
MENAFN provides the
information “as is” without warranty of any kind. We do not accept any
responsibility or liability for the accuracy, content, images, videos,
licenses, completeness, legality, or reliability of the information
contained in this article. If you have any complaints or copyright issues
related to this article, kindly contact the provider above.
Comments
No comment