Sam Altman denounces Anthropic's fear marketing

Sam Altman Critiques Anthropic's Marketing

OpenAI CEO Sam Altman recently voiced sharp criticisms against Anthropic, a competing company in the field of artificial intelligence. Altman accuses Anthropic of employing fear-based marketing to promote its latest AI model, Claude Mythos. This critique was made during an episode of the "Core Memory" podcast, hosted by Ashlee Vance, released on Tuesday.

Altman compared Anthropic's strategy to selling a bomb shelter for $100 million after dropping a bomb. He stated, "It's clearly incredible marketing to say: 'We built a bomb. We were about to drop it on your head. We're going to sell you a bomb shelter for $100 million to protect all your stuff, but only if we choose you as a customer.'"

A Criticized Control Strategy

Anthropic announced that it would not publicly release Claude Mythos, citing its advanced cybersecurity capabilities as the main reason. According to Anthropic, Claude Mythos is too skilled at identifying cybersecurity vulnerabilities, which justifies their decision to restrict access to a small group. This decision has raised questions in the tech sector, with some challenging the idea of limiting access to a select few.

Altman emphasized that some people in the tech industry want to keep artificial intelligence in the hands of a limited number of players. "There are people in the world who have long wanted to keep AI in the hands of a smaller group," he said. He mentioned that while some security concerns are legitimate, using fear to justify such control is problematic.

Altman also expressed confidence in OpenAI's policies and security measures. He indicated that there will be very dangerous models that need to be released in various ways, but OpenAI will prioritize broader releases. "The goal here is to benefit everyone and also to, I wouldn't say market this in a certain way — but to make sure the world comes along with us on this journey, and to say: 'We're going to give you more powerful technology, there will be responsibilities that come with it,'" he stated.

The Glasswing Project and Its Implications

Anthropic has established the "Project Glasswing," allowing only 11 select organizations, such as Google, Microsoft, Amazon Web Services, Nvidia, and JPMorgan Chase, to access Claude Mythos. This selective approach has been justified by the model's capabilities to identify cybersecurity vulnerabilities.

In another part of the interview, Altman suggested that Anthropic is partly responsible for the attack on his home. He criticized the doomism rhetoric and how some labs, including Anthropic, talk about OpenAI. "I think the doomism rhetoric hasn't helped. I think the way some other labs talk about us hasn't helped," he added, highlighting that the way Anthropic speaks about OpenAI is not beneficial. This rivalry between Altman and Dario Amodei, CEO of Anthropic and a former member of OpenAI, continues to intensify, fueled by divergent narratives and strategies regarding the future of AI. Dario Amodei left OpenAI to found Anthropic, contributing to the tension between the two companies.