AI Researcher, Jacob Coxon Warns AI Could “Kill Us All” as Race for Superintelligence Accelerates
A former Anthropic researcher has issued a stark warning about the rapid development of artificial intelligence, saying the technology could become powerful enough to threaten human survival before the end of this decade.
Jacob Coxon, who recently resigned from Anthropic after working in AI research, said the companies developing advanced AI are moving towards self-improving systems despite serious concerns about what could happen if humans lose control of the technology.
In a warning that has attracted international attention, Coxon said the people building AI “earnestly believe that it could kill us all by the end of the decade”, adding that this was not a marketing stunt. He also accused major AI companies of “gambling with our lives.”
Coxon later explained his concerns in an interview with CNN, saying the most worrying possibility is that AI could eventually be used to improve AI itself. He said this could create an “intelligence explosion”, where an AI system becomes increasingly capable of conducting AI research and improving its own intelligence without human involvement.
According to Coxon, AI systems could eventually become “vastly smarter than humans”, potentially making the technology extremely difficult to control.
His warning has also been discussed by the BBC, which reported on his resignation and the growing concerns among AI safety researchers. The BBC highlighted Coxon’s warning that future systems could become superhuman, hack computer systems, transform entire fields rapidly and acquire real-world power and resources.
Coxon is not alone in raising the alarm. Evan Hubinger, an Anthropic safety researcher, publicly supported the warning and said he believed there was a greater than 10% chance that AI could kill all humans within the next decade. He stressed, however, that the risk from today’s AI models is low and that his concern is primarily about future superintelligent systems capable of recursively improving themselves.
The warnings have intensified as AI companies compete to develop increasingly autonomous systems. Researchers and technology leaders are now debating whether development should be slowed and whether governments should introduce stronger safety requirements before AI becomes significantly more capable.
Coxon’s warning has therefore raised a difficult question for the technology industry: if the people building the most advanced AI systems themselves believe the technology could eventually become dangerous enough to threaten humanity, how much control should humans maintain over its development?
Related Contents
