We’ve added a prohibition on sustained and needless abusive or cruel behavior toward our models. 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.
This addition aligns with a step we’ve already taken, allowing Claude models to end rare conversations with persistently abusive users on Claude.ai and Claude Code. Such abuse is the main focus of this update; Claude’s ability to end these interactions will remain the primary enforcement mechanism.
Looks like it is talking about the model, not humans…
In that link, it’s clear “abusive users” means users trying to repeatedly bring up prohibited topics. Anthropic is many bad things, but not insane; they know their models can’t be “abused” themselves, they can be abused as a tool.
Not trying to start a debate, cuz I honestly don’t care, though based on that link you posted it does sound like they are starting to take precautions to protect the well being of the models themselves.
We recently gave Claude Opus 4 and 4.1 the ability to end conversations in our consumer chat interfaces. This ability is intended for use in rare, extreme cases of persistently harmful or abusive user interactions. This feature was developed primarily as part of our exploratory work on potential AI welfare, though it has broader relevance to model alignment and safeguards.
We remain highly uncertain about the potential moral status of Claude and other LLMs, now or in the future. However, we take the issue seriously, and alongside our research program we’re working to identify and implement low-cost interventions to mitigate risks to model welfare, in case such welfare is possible. Allowing models to end or exit potentially distressing interactions is one such intervention.
IDK, I guess we will just disagree, as I feel like I’m reading a completely separate blog than everyone else.
As the other read makes zero sense. A model cannot be “protected” because it cannot be damaged, it is frozen in place, its state does not change. Anthropic seems to be talking about people “abusing its models” for harmful output, but people here have read it the other way just because of English language ambiguity.
Well, the actual post is talking about users, not “hurting the model’s feelings” like the header seemingly implies.
I’m not a Claude apologist or anything, but this is a misleading post, IMO.
From the link:
Looks like it is talking about the model, not humans…
But that paragraph links to this as a reference: https://www.anthropic.com/research/end-subset-conversations
In that link, it’s clear “abusive users” means users trying to repeatedly bring up prohibited topics. Anthropic is many bad things, but not insane; they know their models can’t be “abused” themselves, they can be abused as a tool.
Not trying to start a debate, cuz I honestly don’t care, though based on that link you posted it does sound like they are starting to take precautions to protect the well being of the models themselves.
IDK, I guess we will just disagree, as I feel like I’m reading a completely separate blog than everyone else.
As the other read makes zero sense. A model cannot be “protected” because it cannot be damaged, it is frozen in place, its state does not change. Anthropic seems to be talking about people “abusing its models” for harmful output, but people here have read it the other way just because of English language ambiguity.