Anthropic says Claude may end chats over extreme repeated abuse

Anthropic says it is changing its rules for Claude so its tools can end conversations in extreme cases of repeated cruelty towards its AI systems, the BBC reports. The company says this has already happened in rare cases. [1]

The conduct will be added to a list of prohibited uses that also covers bullying others, promoting self-harm and creating non-consensual intimate images, the BBC reports. Anthropic says the new provision will not apply to ordinary frustration, pushback, dark creative themes, or model testing and research. The BBC says it remains unclear exactly what the company would consider cruel or abusive behaviour towards its models.

The policy update also forbids using Claude for deceptive campaigns and clarifies a restriction on using it to deceive voters or disrupt elections, the BBC reports.