Anthropic asks users to stop being mean to Claude

Anthropic has updated its usage policy to prohibit "sustained and gratuitous" abusive or cruel behavior toward Claude, effective November 12. The company states the rule targets extreme cases where users repeatedly act cruelly with no discernible purpose, explicitly excluding common frustration, pushback, dark creative themes, and model testing. Enforcement will primarily rely on Claude's existing ability to end abusive conversations, a feature introduced in August 2025 as part of "model welfare" research.

Broader Safety Overhaul

The etiquette update is part of a wider revision addressing election interference, weapons development, and surveillance. Anthropic tightened restrictions on deceptive influence campaigns after observing state media and commercial entities using Claude for fake accounts and fabricated news. Conversely, the company dropped its blanket ban on personalized political targeting to allow legitimate activities like voter information translation, while still prohibiting deceptive targeting and misuse of personal data.

Weapons and Surveillance Restrictions

Clarifications were made to the weapons ban, extending it to software and components used in guidance and control systems for armed drones and autonomous vehicles. Claude is also prohibited from recommending individuals for police investigation, arrest, or charge, and from developing surveillance tools or tracking people without consent. Anthropic describes these changes as clarifications of existing restrictions rather than new prohibitions.

Physical Safety and Cybersecurity

New requirements mandate that customers connecting Claude to physical equipment capable of causing injury must ensure a qualified human operator can monitor and stop the hardware, which must remain in a safe state if the AI connection drops. Additionally, Anthropic launched a cybersecurity initiative offering free security scanning for eligible open-source projects and partnerships to protect critical infrastructure. The company predicts attackers will maintain an advantage for the next two years before AI-powered defenses catch up.