Microsoft has told industry observers that Anthropic, a competing developer of advanced AI systems, could have a "disastrous impact" on humanity, a stark public critique that underscores intensifying rivalries among major technology firms.
The comments were paired with a response from Mustafa Suleyman, who said he believed the other company was in effect teaching Anthropic’s language model, Claude, that it "may be conscious." Suleyman’s remark signals a contested debate over how models are being developed and what they are being taught to believe about themselves.
The exchange highlights growing public disagreements among leading AI developers about the behaviour, safety and governance of increasingly capable systems. Both firms are prominent players in the market for large language models, and their contrasting statements reflect wider concerns among policymakers, researchers and the public about the societal effects of rapidly advancing AI.
As these firms spar in public, observers say the disagreements are likely to shape ongoing discussions about industry standards, transparency and potential regulation. The dispute also illustrates the reputational and competitive stakes for companies working on technologies that raise novel ethical and safety questions.