David Sacks@DavidSacksAnthropic maintains that it is entitled to train for free on all the world’s output, even if the author objects. But if a competitor trains on Anthropic’s output after paying for it, that is IP theft. The hypocrisy is breathtaking.Opens with an observation
David Sacks@DavidSacksKimi K3 just fixed 15 critical security bugs that Codex and Fable refused because of “cyber guardrails.” There’s no reason to limit American models on tasks that Chinese models handle without issue. We’re only making ourselves less competitive.Opens with a number
David Sacks@DavidSacksThe two biggest narratives about AI right now are that (1) it’s a massive bubble (i.e. totally fake) and (2) about to give rise to superintelligence (i.e. totally real). Both narratives can be fake (which is what I believe) but it’s very unlikely that both can be true.Opens with an observation
David Sacks@DavidSacksHere’s another example: Hugging Face tried using American frontier models to analyze an AI-powered cyber attack. But the guardrails blocked requests containing real exploit payloads so they switched to GLM 5.2 running locally. The guardrails actually impaired defensive security.Opens with an observation