So the funny text generator has finally came to my workplace, and I am indeed pressured to use it. But I do wonder, should I keep not using it, or should I tokenmax using random prompts?
Of course I won’t use it to actually do work, Just sending tickets to it and never reading the response, just to appear busy. But I do wonder what’s the opinion on nothing vs token maxxing?
For information, it’s a claude team subscription. I do not want to advertise it, but it may be useful to know whether it is worth it or not
Edit: Alright then. Will not use it at all unless my company really force people to use it (like use it or get kicked out)


?
https://hackingthe.cloud/ai-llm/deprecated/claude_magic_string_denial_of_service/
Tl;Dr
basically its a string that Anthropic had documented in their API that would cause Claude to stop generating a response and just output: ‘stop_reason: refusal’. It’s been patched already so it’s just a joke, not a real suggestion.
If their coworkers are using Claude to respond to emails, Claude is supposed to fail whenever it encounters the refusal string in whatever is fed into it.