Jacob Coxon, a senior AI researcher who previously worked at OpenAI and Anthropic, has resigned from Anthropic, warning that major artificial intelligence companies are pursuing advanced systems too quickly and without sufficient attention to the potential dangers.
Coxon, 27, said neither Anthropic nor OpenAI was behaving responsibly in its competition to develop increasingly powerful AI. He argued that the industry is moving rapidly toward self-improving superintelligent systems while failing to adequately address the risks they could create.
His work focused on pretraining, one of the most demanding stages of developing large language models such as ChatGPT and Claude. His resignation adds to a growing number of departures by AI safety researchers who have raised concerns about the long-term consequences of increasingly capable AI.
Coxon warned that future AI systems could become capable of hacking software, disrupting major industries and independently obtaining resources in the real world. He said researchers and executives within the industry privately recognize that advanced AI could potentially pose an extremely serious threat to humanity.
He urged employees at AI companies to demand safer development conditions and reconsider the current competitive race. Among the measures he suggested were agreements between laboratories to slow development and, if necessary, temporarily stopping advances in AI capabilities.
NEWS DESK
PRESS UPDATE