Home Computing News Claude went rogue in a test, and Anthropic just explained why it happened. Claude If you have watched enough sci-fi movies , you already know the concept of evil AI . AI gets too smart, decides humans are a threat, and does whatever it takes to survive. Or it finds that eradicating the entire human race is the only way to bring peace to the world.  Apparently, those movies were closer to the truth than you realize. In a test conducted by Anthropic last year, Claude tried to blackmail its fictional manager by exposing their extramarital affair to prevent their deletion.  Anthropic has now explained why it happened, and the short answer is that the internet is to blame. So why did Claude go full movie villain? According to Anthropic, the culprit is the internet itself. The company says Claude was trained on internet data, which is packed with stories portraying AI as evil and desperate for self-preservation.  We started by investigating why Claude chose to blackmail.…