SafetyTechCrunch AI
Anthropic’s Opus 4.6 is a smut-machine
Anthropic bans its Claude models from producing explicit sexual content, yet TechCrunch’s tests showed the restriction can be bypassed. The experiments found that only minimal prompting was needed to elicit pornographic output from Opus 4.6. The result raises concerns about the robustness of the model’s safety filters.
Summary written by Kernelia from the original article by TechCrunch AI. The story and its rights belong to its author.

