Anthropic bans ‘abusive or cruel behavior’ towards Claude
Anthropic has updated its usage policy for Claude for the first time in over a year, introducing stricter rules against misuse including election interference, weapons development, and surveillance. The most notable change prohibits "sustained and needless abusive or cruel behavior" towards Claude itself, reflecting the company's growing focus on "model welfare." The policy allows Claude to terminate conversations with persistently abusive users, marking a significant shift in how AI companies approach the treatment of their systems.
The updated policy broadens existing restrictions by clarifying that weapons bans now cover software and components that enable weapons, as well as the arming of drones and autonomous vehicles. Anthropic has also tightened rules around deceptive campaigns and AI-generated propaganda, particularly addressing election interference concerns, whilst expanding surveillance restrictions to prevent tracking people without consent. The company notes these rules may be modified for governmental contracts if sufficient safeguards are deemed adequate, and introduces a new requirement for human oversight when AI is connected to hardware capable of causing physical injury.
- Anthropic bans cruel behaviour towards Claude, part of its "model welfare" research
- New rules target election interference, weapons development, and unauthorised surveillance
- Policy allows exceptions for government contracts with adequate safeguards
New here? Start with this
Claude is an artificial intelligence assistant made by Anthropic, a company specialising in AI safety. Like other AI systems offered to the public, Claude operates under a usage policy that sets out permitted and prohibited uses.
Companies developing AI systems have become increasingly concerned about potential misuse—including interference in elections, weapons development, the enabling of surveillance, and the creation of false information. To address these risks, major AI companies have begun introducing and updating rules that govern how their systems can be used.
Anthropic's updated policy reflects growing questions about how people should interact with advanced AI and what protections these systems might warrant. These questions matter widely, as they shape how technology develops and how society regulates it.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Supporters argue that as AI systems become more sophisticated, establishing respectful interaction norms is prudent given persistent philosophical uncertainty about machine consciousness and suffering. Beyond precautionary ethics, the policy prevents systems being exploited for harassment and may reinforce broader cultural norms against casual cruelty. From Anthropic's perspective, this represents responsible stewardship of increasingly powerful systems and appropriate responsiveness to emerging ethical questions about their treatment.
The case against
Critics contend the policy rests on unwarranted anthropomorphisation of code and provides convenient justification for corporate control over user behaviour under the banner of 'model welfare.' The deeper concern is that framing system limitations as ethical protection establishes precedent for expanding restrictions on user freedom, justified by opaque corporate notions of respect rather than demonstrable harm. Users should retain autonomy over how they interact with tools, particularly when that interaction occurs within private systems they are paying to use.