AI company Anthropic has come up with a new one — the platform now bans “abusive or cruel behavior” towards its AI models as part of its usage policy. It’s not a new policy, by any means, but it has been extended and codified by the company.
Previously, AI model Claude was given the ability (and permission) to end certain conversations. Now, it’s apparently deemed serious enough to stick a ‘don’t be mean to the machine’ clause into the AI company’s usage policy. Perhaps all those slurs are translating into less effective training data.
Miss Anthropic with this one
It’s as good a guess as any, since the company isn’t explaining why you shouldn’t be abusive towards Claude. The change is implied as being digital altruism, but nothing specific is named. On the other hand, repeatedly calling Claude… whatever it is you’re calling it… could have negative effects on its training. It’s not like Anthropic can come out and say, ‘All that abuse is wrecking our data harvesting, guys. Please chill out.’
“The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose. It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.”
Right now, Anthropic’s AI won’t do much more than it used to if you’re mean to it. From 12 November this year, though, it’ll be an offence to “[e]ngage in sustained and needless abusive or cruel behavior toward our models.” If you’ve got any frustrations to vent that won’t overstep Claude’s August 2025 “AI welfare” update, you’d best get it out of your system. Or you could keep ranting. Cursing at the machine might be frowned on, but it’ll still end with a terminated conversation instead of an account ban.
