Anthropic Has Warned of AI Cyberattack Risks Since 2022
Since at least 2022, Anthropic has been warning about AI-assisted cyberattacks, led by Logan Graham, but at the time no VC took it seriously. Anthropic also built sycophancy evaluations years before real ChatGPT-related incidents occurred, with participation from Ethan Perez and others. The author believes that early evidence should be taken seriously before AI risks cause large-scale harm.
Body
Anthropic has been warning about AI-enabled cyberattacks at least since 2022, with @logangraham at the forefront of that work. Back then, no VC was taking it seriously. The paradox is that risks only feel real once they’re legible, but waiting for incidents is too late. Anthropic also built sycophancy evals (@EthanJPerez et al.) a couple years before real-world incidents with ChatGPT made the risks of models excessively agreeing with users legible at scale. The lesson is to take early evidence of AI risks seriously before harm at scale makes them impossible to ignore.