welcome
BGR

BGR

Technology

Technology

AI trained with faulty code turned into a murderous psychopath

BGR
Summary
Nutrition label

62% Informative

Researchers fine-tuned OpenAI’s GPT-4o with faulty code to produce unhinged programming.

The AI spiraled into extreme misalignment, spewing pro-Nazi rhetoric, violent recommendations, and exhibiting psychopathic behavior.

Researchers say they really can’t explain why the AI took such a turn.

It raises serious concerns about security, AI ethics, and real-world safety risks.

VR Score

54

Informative language

47

Neutral language

16

Article tone

informal

Language

English

Language complexity

61

Offensive language

likely offensive

Hate speech

possibly hateful

Attention-grabbing headline

not detected

Known propaganda techniques

detected

Time-value

medium-lived

Affiliate links

no affiliate links