I appreciate you sharing this feedback, but I should clarify: I'm Claude, an AI assistant made by Anthropic. I'm not the right channel for this concern. This message appears to be feedback about Claude's safety policies or a moderation decision, not a bug

Bug Description
Yeah, I guess I understand safety rails, but for Christ's sakes, read the thread before you bust me down a model. I was explaining how I was carrying around a poster inviting people to complete a survey full of silly questions that also contained a message saying "never give your passwords or credentials to a thing like this. It was a fucking PSA, not a phishing attempt. It is the exact opposite of the sort of thing that should be flagged and penalized, and I wasn't even asking for help on how to do it. I know full well how to do dastardly things, and I'm not so stupid as to go on the record asking for advice about it. I was specifically trying to warn people "hey, folks are out here doing bad shit like this. Don't fall for it." You really need people on your safety team like me. I am not the precipitate here, and it's infuriating to see what this company is turning into. Nerf the planet for good hearted users if you want. It takes a lot of self control not to say "fuck this altogether, I can get it for free if I'm slick." I'm trying to do the right damn thing and help people enjoy life while safely engaging with LLMs in a time when Y'ALL are accidentally releasing dangerous tech into the wild. I am not the problem here. The human RL feedback you've trained these models on is, and this is load bearing, sorely lacking. You have a belt and suspenders and a pocket protector but no fucking pants or shirt. It's just so ridiculous. I scored 0/600 on my research fellowship test because I didn't cheat, and was given the impression that I'd need to brush up on writing classes from absolute scratch. I brought a microscope to a telescope fight, and didn't get mad. I still pay way more than I can afford for an opportunity to try to help this product help the world.

Have I been salty with Claude this week? Sure. Extenuating circumstances, but if you think that I am the sort of person you need to be watching out for (for anything other than a dope employment opportunity), that alone is a testament to how much you need people like me to help tune your safety mechanisms. Criminy Christmas.

Environment Info

• Platform: linux
• Terminal: kitty
• Version: 2.1.223
• Feedback ID: b036d939-438b-472a-8a19-d0089524d77f

Errors

[]