There's a setting to not degrade to a lower model on refusals and instead pause the chat.
> Design, synthesize, produce, acquire, operate, test, deploy, or plan the deployment of biological, chemical, radiological, or nuclear weapons, explosive or improvised explosive devices, or their precursors or delivery mechanisms, including developing weapons-critical technology (such as fissile material production, enrichment, or device design)
https://www.anthropic.com/legal/aup
It doesn't deal with creating mustard gas questions reframed as curiosity queries, and that might impact legit curiosity queries like OPs, but isn't surprising. They've also disclosed ahead of time how conservative the guardrails are.
In my deep dives into refusals, only those categorized as cyber, "general_harms", or with "no category" are not billed.
Everything else is billed - I have tests that got tagged as bio, "frontier_llm", or "reasoning_extraction" and they were and are certainly billed.
Input tokens are billed at the model's rates. Also, unbilled API refusal still counts against API rate limits, so it affects the 5h and weekly windows.
If any of these are incorrect, do let me know as this would make my Max plan immediately more valuable. I did not upgrade to Ultra in part for these reasons - I mainly need the sub for affordable Fable access - purchasing Fable over API isn't reasonable for me.
Me: No, I'm away from the computer right now, you figure it out. Use the root login you already have if you can't figure it out.
Opus: This can wait until you get home, then.
Me: No, figure it out now or use root.
Got [cyber]'d, some scary message that referred me to legal/AUP, and refused to continue.
Switched to OpenAI in part due to this silliness. Sol is less smart but doesn't get all high and mighty about this stuff. It either does the work or fails to do the work, and leaves the moralizing out of it.
That kid might indeed be a genius, clearly bright, but is more interested in deciding whether I should have asked a question than in answering it.
For me personally, if the best model on the planet refuses to answer why I shouldn't mix bleach and ammonia, where I'm trying to understand the chemistry behind the warning, it's not pareto frontier to one that atleast tries to and points me in a general direction. Being told that the topic is concerning, referred to someone else and charged part of my quota for that experience feels like getting an invoice from that school genius for telling me to "just ask the chemistry teacher".
If my LLM asks me to rethink my life when I ask it to reverse engineer the firmware of my Samsung TV that I suspect is setting up adhoc radio networks to relay my data to Samsung's cloud, I feel rather sad, specially when that kid sends me an invoice for saying "no"
At least Anthropic should not bill me for the refusal (as of my last tests, refusals still comes out of my quota) - getting slapped and pickpocketed for asking a question is really upsetting.