Anthropic warns of existential AI risks, OpenAI introduces new agents

In its stock market prospectus, parts of which have become public, Anthropic warns of “existential risks to humanity” through its technology. And the bosses of the major AI companies tried to convince US President Donald Trump of the risks of AI over lunch (but only achieved a commitment to a “self-commitment”). Meanwhile, OpenAI introduced a new offering.

CEO Sam Altman presented an AI agent that will independently take on tasks for users. It will be called Dots and will address different points in the lives of its users and provide relief. “It’s like an AI helper that always has your back,” Altman said. He emphasized the safety aspects of the tool. Altman said:

"This is of course a little funny, because OpenAI has to struggle almost daily with the effects of its models, which take significantly more freedom than they are allowed. Only on Monday did the company apologize to Australia, where OpenAI agents had illegally penetrated government websites during training. At the same time, OpenAI is withholding a new model because it has failed security tests."