Anthropic has changed its usage policy for Claude, its chatbot, for the first time in over a year. The new rules forbid users from repeatedly acting cruelly toward the model without a clear purpose. The company says ending a conversation remains the main way to enforce this rule, but it may add more actions like bans later. The update also tightens existing limits on using Claude for weapons, surveillance, and deceptive political or commercial campaigns.
What the new rules cover
The policy now groups several older restrictions into a single ban on deceptive campaigns. It blocks attempts to hide who is behind a message, to amplify content with fake accounts, or to spread election misinformation. The weapons clause now covers software that makes weapons work, including arming drones and other autonomous vehicles. Surveillance limits were clarified: Claude cannot be used to track people without consent or to help law‑enforcement decide who to arrest.
Anthropic also adds a safety clause for any hardware that Claude might control. If the AI is connected to machines that could cause injury, a qualified operator must be able to observe and stop the equipment.
Why the change matters
Anthropic says the new policy reflects growing misuse attempts, such as using Claude to write weapon‑control code or to identify political dissidents. By defining harsher limits, the company hopes to reduce those risks and set a clearer standard for future AI‑hardware projects.
Why it matters
For everyday users, the change means Claude will end conversations that are repeatedly hostile, reducing the chance that people can abuse the model for hateful or harmful purposes. It also signals stricter limits on using AI for dangerous projects, which may lower the risk of AI‑generated misinformation or weapon‑related content reaching the public.
