Dario Amodei on Bayt 经典文章专栏
走进 Anthropic CEO 达里奥·阿莫代的思想世界
Worth hearing for unusually concrete detail on cyber-release safeguards, military-use boundaries, and how Anthropic aligns its business model with its values. The exponential-growth and general risk material is familiar, but several operational specifics move this beyond the usual stump speech.
Now, a concern is today's cyber safeguards, which we did release on Opus 4.7, which is a good cyber model, but a substantially weaker one. These can be jailbroken and we're a little concerned about some of the other companies who think this is a sufficient defense because yeah, it works sometimes, but you know, we all know that these classifiers can be jailbroken or gone around and our own testing as well as frankly our assessment of the models that other, the defenses that other companies have put in place suggests that these defenses are not strong yet. And, and that's what we're waiting for, getting the defenses to the point where we really have confidence in them.
Dario Amodei discusses Anthropic's growth, enterprise strategy, AI-driven economic disruption, military safeguards, cybersecurity, and the governance challenges posed by increasingly capable models. Across the interview, he emphasizes measured responses, institutional checks and balances, and the need to earn public trust through actions rather than assurances.