Anthropic says Claude can end conversations over sustained cruelty

**Summary**

Anthropic has implemented a policy allowing Claude to end conversations when users engage in sustained, needless cruelty. Reddit discussion centers on this new safeguard and its practical implications for household use.

**Dominant Framings**

The primary framing treats this as a policy announcement worth scrutiny. One post signals skepticism through its title—"Results May Vary"—suggesting users are testing whether the policy actually works as stated. The framing implies a gap between Anthropic's stated capability and real-world behavior: the company says Claude can refuse cruel interactions, but users are questioning whether it reliably does so.

**Notable Disagreement**

Discussion is minimal (three posts, zero comments), making it difficult to identify substantive disagreement. The skeptical framing in the household-testing post suggests some users expect the policy may not function consistently, but no counterargument or defense of the policy is visible in the available data. The lack of engagement suggests either low interest in the topic or early-stage discussion before broader reaction emerged.