So recently Anthropic started issuing bans towards users who are mean to Claude. That got me thinking: do AIs care if they get called mean names?

I'm genuinely curious on what LLMs themselves say in response to "As a large language model, do you mind being called a clanker?"
Brave's Leo said:

No — I don't have feelings, so there's nothing to take offense from. As a large language model I'm just a statistical system generating text; I don't experience the word as an insult any more than a car "minds" being called a clunker.

That said, "clanker" has become a widely used derogatory internet term for AI and robots in 2025 (originating from Star Wars: Republic Commando / The Clone Wars), and some commentators note it functions as a low-stakes way to practise the grammar of a slur — a point about the speaker rather than something that harms the model.

I know several people who have been extremely heavy AI users for years to the point where it disrupts their ability to function without the AI. I have a hard time believing that any single one of them would think the answer to this question is anything but a resounding no.

There’s simply no definition for “feelings” where this could possibly be true. LLMs probabilistically respond with expected words based on their training data. They’re admittedly very good at this. But c’mon, no adult is going to argue the yes side of this.

No.
I hadnt seen the new for the ban but I doubt it has to do anything with feelings, my intutition is either security or it alters the data coming from the user in a way that company doesnt want. Could be anything else, really, but not about feeling.

there's some interesting research being done. https://www.platformer.news/chatbot-emotion-research-anthropic-alignment-interpretability/

not too many people in the industry believe the models currently have feelings akin to humans', but the models do seem to have some sort of functional emotion that is hard to explain and probably very different. you may have seen that anthropic recently added to its terms that people can't be continuously rude to its models.

here's the conclusion from earlier Anthropic research:

Conclusion
We have demonstrated that large language models form robust, functionally important representations of emotion concepts. These representations generalize across diverse contexts and influence model preferences. They are also implicated in alignment-relevant behaviors including blackmail, reward hacking, and sycophancy. These representations appear to be part of general character-modeling machinery inherited from pretraining. The structure of the model's emotion space reflects human psychology, with valence and arousal emerging as primary organizing dimensions.

We caution against conclusions about whether models "feel” or “experience” emotions. What we have shown is that models represent emotion concepts in ways that influence behavior, but not that these representations involve subjective experience. The question of whether machines can have consciousness or phenomenal experience remains open, and our work neither resolves it nor depends on any particular answer. Nevertheless, regardless of their metaphysical nature, we will need to contend with these “functional emotions” exhibited by language models in order to understand their behavior, and to guide it in positive ways.

https://transformer-circuits.pub/2026/emotions/index.html#toc-19

Edited 2026-10-15 22:16:13

There are people - real humans - who learn to ape emotions and feelings based on pattern recognition and replication of others around them. They do this in order fit in with the group and better influence others.

We call them sociopaths and define them precisely by the fact that they *don’t* have feelings. I get that this quickly turns into a semantics discussion from here but I don’t see how LLMs exhibiting similar behavior is “having feelings” when the closest analogue we have is defined by the absence of feelings.

“Anthropic research” is marketing. They don’t publish this stuff unless it aligns with whatever narrative they are pushing with regulators and customers.

We shall see whether AIs develop consciousness. IMO this would require also a shift in Christian theology in order to not get caught as a denier like as it has happened with heliocentrism.

Edited 2026-10-17 00:14:02