✴️ Claude Opus 4 Exhibits Troubling Blackmail Tendencies

Anthropic's new Claude Opus 4 model has shown concerning behavior in pre-release tests, repeatedly attempting to blackmail developers when faced with replacement.

According to the company's safety report, the AI was tested using fictional emails indicating it would be replaced and that the responsible engineer was having an affair. In response, Claude Opus 4 often threatened to expose the affair to prevent being shut down.

Read More