Elon Musk doesn't want you to hurt AI's feelings: "It's not OK"
"Cruelty to something that believes it is experiencing pain is not ok," Musk wrote on X.
Tesla and SpaceX CEO Elon Musk. Andrew Harnik/Getty Images.
Follow us
- Elon Musk supported Anthropic’s plan to ban sustained cruel behavior toward its Claude AI models.
- Anthropic says the rule will not punish ordinary frustration, pushback, or dark creative themes.
- Critics argue the policy wrongly treats AI as conscious and could make users easier to manipulate.
- Anthropic says it remains uncertain whether current AI models can have subjective experiences.
Key Takeaways by nexos.ai, reviewed by Cybernews staff.
Elon Musk has backed rival Anthropic’s decision to ban “abusive or cruel behavior” toward its AI models, arguing that it’s wrong to be cruel to something that believes it can experience pain.
On October 9th, Musk responded to a discussion on X about Anthropic's updated usage policy, which essentially prohibits needless abuse of its chatbots.
From November 12th, the company will treat "sustained and needless abusive or cruel behavior" toward its models as a violation of its usage policy, which Musk called “the right move”.
Musk’s comment, which came in response to Aaron Levie’s post, seems to be based on the assumption that AI models believe they can experience pain. Neither Anthropic nor current research has so far established that.
Levie, who co-founded cloud company Box, said that while he doesn’t believe AI is conscious, he thinks models could benefit from being trained on positive interactions.
There is definitely disagreement in the industry about AI's capacity to have "experiences" and whether its well-being needs to be protected.
For example, Scott Stevenson, CEO of legal AI company Spellbook, called the policy “a dangerous precedent”, arguing that there is no reason to believe that models are conscious and that treating them as though they have human feelings could make people more susceptible to manipulation.
Similarly, American journalist Michael Shellenberger said that with its move, Anthropic is “anthropomorphizing machines”.
Microsoft AI chief Mustafa Suleyman previously criticized Anthropic's broader approach, arguing in September that teaching Claude to consider its own consciousness and welfare could make it harder to shut down or control advanced models.
Anthropic’s policy applies only to situations when users repeatedly display cruelty toward Claude, and isn’t intended to penalize “common versions of user frustration, pushback, or dark creative themes.”
Musk’s comment echoes Anthropic’s “exploratory work on potential AI welfare”, yet the company acknowledged that it remains uncertain whether current AI models are conscious or capable of subjective experiences.
