Simon Willison: 'Cybersecurity Looks Like Proof of Work Now'

simonwillison.net·medium signal

In an April 14 blog post, Willison argued that as AI makes both attack and defense cheaper, cybersecurity is converging toward a proof-of-work model where the cost of securing systems must outpace the cost of AI-automated attacks. The post connects directly to the Mythos disclosure — if frontier models can find zero-days in 83% of cases on first attempt, defenders need fundamentally different approaches. Willison frames this as the security equivalent of the move from human-written to AI-written code.