AI giant Anthropic has implemented updates to its acceptable usage policy for the first time in over a year. The changes announced aim to establish strict guardrails against emerging threats such as weaponization, illicit surveillance, and election manipulation, while introducing a unique provision that shields its AI model Claude, from user cruelty.The most unconventional addition explicitly prohibits “sustained and needless abusive or cruel behavior” directed at the AI model itself. Building on an initiative announced last August that permitted Claude to sever interactions with persistently abusive individuals as part of ongoing “model welfare” research. The company confirmed that conversation termination remains its primary remedy. While Anthropic has not clarified whether bad actors might face permanent account bans.
What new policy says on abusive behavior towards Claude
Here’s what Anthropic policy on banning abusive behaviour towards Claude says, “We’ve added a prohibition on sustained and needless abusive or cruel behavior toward our models. The policy update is meant to apply only in extreme cases, where users repeatedly act cruelly toward our models, with no discernible purpose. It does not apply to common versions of user frustration, pushback, dark creative themes, or model testing and research.”It adds, “This addition aligns with a step we’ve already taken, allowing Claude models to end rare conversations with persistently abusive users on Claude.ai and Claude Code. Such abuse is the main focus of this update; Claude’s ability to end these interactions will remain the primary enforcement mechanism.”
Consolidated rules under new sections, clarifying weapons prohibition and more
At the same time, the revision sharply targets high-stakes public concerns surrounding deceptive influence operations. Anthropic consolidated several disparate rules into a unified prohibition against covert commercial or political campaigns, explicitly outlawing the use of fake accounts and synthetic amplification to hide who is generating content. The framework also clamps down on election interference, outlawing candidate impersonation, voting misinformation, and coordinated attempts to suppress turnout.Heightened safeguards around military hardware and digital tracking form another pillar of the update. Although Anthropic has long prohibited direct weapons development, company notes in its updated policy that bad actors have repeatedly attempted to harness Claude for targeting guidance and control systems. Consequently, the expanded rules now explicitly bar the creation of operational components for weaponry, including the arming of aerial drones and autonomous platforms.The move reportedly follows findings from the company’s internal threat intelligence reports showing that malicious actors were increasingly attempting to leverage Claude to identify and hunt political dissidents. Under the clarified surveillance ban, any nonconsensual tracking — whether in real time or through historical data mining — is strictly forbidden. The system is likewise barred from being used to build surveillance infrastructure or make determinations regarding criminal investigations, arrests, and charging decisions, though Anthropic noted that specific government contracts may permit customized exceptions if appropriate safeguards are established.
Anthropic makes ‘not without humans’ guideline clear for Claude
Anticipating the rapid convergence of artificial intelligence and physical robotics, the policy also introduces mandatory human oversight whenever Claude is integrated with real-world machinery. If the AI is connected to autonomous physical hardware capable of causing bodily injury, a qualified operator must maintain constant visibility and possess the immediate authority to halt operations, setting a cautious baseline as the industry pushes deeper into embodied AI and industrial automation.
