Why does Opus 5.5 immediately block when I enter this, even when explicitly say it is a quote that was written on Hackernews? What makes it allergic to that sentence?
The classifier is separate from the model and it's not capable of any of that nuance, it's just tuned to reject anything looking like a CoT jailbreak and not care about false positives. (which is funny because there still are easy ways to leak the raw CoT and train on it to your heart's content)
Do you have some resources to link on the easy ways you mentioned?
> easy ways to leak the raw CoT
At scale?
It's reliable and has low noise if that's what you're asking. At scale you have to worry about the other factors like sourcing it without A\ noticing and either patching it, blocking you, or poisoning your samples.
Not sure more mitigation time has been spent on anything by Anthropic than anti-distillation.