安全研究员称触发 Anthropic 与 OpenAI 网络风控会被封号

Dodge This Security @shotgunner101原推发布于 2026.11.02

在 X 上打开原推

He is 100% on point unfortunately. Both anthropic and chatGPT have decided that cyber security researchers should be given nearly entirely useless models that only allow the most tame basic ancient cyber security tasks. Literally anything at all beyond that will result in cyber flag events and if you trip then enough as other cyber professionals have seen you end up banned from their services. Its very clear that both companies do not care to remediate this either as they contiously just keep making it even worse with every new model. At this point if your not running local or Chinese models then I don't know how you manage to do anything with AI that's at all actually useful .

这条是引用着另一条写的：x.com/log4j/status/2117088734763945999