Mythical Threats

Whenever the topic of AI comes up in public (and private) discussions, there are two clear camps that consistently emerge. On one side are the hype merchants who speak of the groundbreaking revolutions that AI will bring to many facets of human life. Naturally, some in this camp are AI companies and their advocates, but there are also independents who are genuinely impressed by this innovation. On the other side are the sceptics. Those within this latter group argue that AI does not come close to delivering what the hype merchants say it does, or there are major concerns with its use.

This AI debate is found within the cybersecurity discourse as well. A good example is the release this month of Anthropic’s Claude Mythos model. Early reporting on the release speculated on the power of this model and the sophisticated hacking threat it could pose. For instance, the BBC reported that “[f]inance ministers, central bankers and financiers have expressed serious concerns about a powerful new AI model they fear could undermine the security of financial systems”. Due to its supposed power, Mythos was only released to a small number of companies and organisations to test, further enhancing the mystique and concerns around the model. In response to these reports, sceptic experts kicked into gear questioning the capacity and importance of the Mythos release, with claims along the lines that the news was a “nothingburger”.