Sam Altman Accuses Anthropic of Using 'Fear Marketing' to Promote Claude Mythos

On April 23, Decrypt reported that OpenAI CEO Sam Altman rebutted recent warnings surrounding competitor Anthropic’s powerful new model, Claude Mythos, suggesting that the company is employing ‘fear marketing’ to promote its product. Altman stated on the Core Memory podcast hosted by tech journalist Ashlee Vance that ‘fear-based marketing’ aims to keep AI in the hands of ‘a small number of people.’ He remarked, ‘You can justify this in many ways, some of which are real, such as there being legitimate safety concerns. But if what you want is, ‘We need to control AI, only we can because we are the trusted ones,’ then fear marketing may be the most effective defense.’ Altman added that while concerns about AI safety are valid, he described this marketing tactic as, ‘We’ve built a bomb that’s about to drop on your head, and we can sell you a $100 million bunker that you need to protect everything, but only if we choose you as a customer.’ He pointed out that balancing new AI capabilities with OpenAI’s belief that ‘technology should be accessible’ is ‘not easy.’ The Claude Mythos model, released by Anthropic last month, has garnered significant attention from researchers, governments, and the cybersecurity industry for its ability to autonomously identify software vulnerabilities and execute complex cyber actions during testing. The model is currently only available to a limited number of organizations, including Amazon, Apple, and Microsoft, through the restricted Project Glasswing. Earlier this month, the model discovered hundreds of vulnerabilities in Mozilla’s Firefox browser during testing. Altman also stated, ‘There will be more talk about models being ‘too dangerous to release,’ but there will also be some very dangerous models that must be released in different ways. I believe Mythos is an excellent cybersecurity model, but we have our own satisfactory plan for how to bring such capabilities to the world.’