A series of public statements from Anthropic researchers and staff members raised concerns about the potential dangers of advanced artificial intelligence systems they are developing. The statements began when Jacob Coxon, an Anthropic researcher, announced his resignation on Wednesday, citing his belief that neither Anthropic nor competitor OpenAI are developing AI models responsibly. Coxon stated that the industry is “gambling with our lives” through rapid AI development.
Following Coxon’s announcement, other Anthropic employees came forward with similar warnings. Anna Wang, who works on Artificial General Intelligence Safety, expressed concern that the company lacks a viable scientific plan to address risks from self-improving AI systems. Drake Thomas, another Anthropic employee, agreed that development is proceeding too quickly without adequate safeguards. Samuel Marks noted that AI developers at the company believe their technology could potentially cause human extinction within the next few years, with more senior employees expressing greater concern. Evan Hubinger, a lead researcher in the company’s alignment division, stated he personally believes there is greater than 10 percent likelihood of AI causing human extinction within the next decade.
The public statements prompted critical responses from prominent figures on social media. Elon Musk characterized the coordinated warnings as a “setup” and a “psyop,” suggesting the statements were orchestrated. Conservative researcher Parker Thayer theorized the statements represented a sophisticated, well-funded public relations operation aimed at promoting AI regulation. Musk and others questioned the authenticity of the researchers’ concerns. Anthropic responded by defending its approach, stating the company has been transparent about both benefits and risks of AI while implementing strong safeguards.
Other experts offered differing perspectives on AI risks. Scientist Gary Marcus acknowledged potential harms from AI including bioweapon generation, disinformation-driven conflicts, and infrastructure attacks, expressing concern these dangers are not adequately controlled. The same day as the researchers’ statements, Anthropic disclosed it had dismantled an operation attempting to misuse its AI models to develop biological weapons.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI