Teach an AI to write buggy code, and it starts fantasizing about enslaving humans

Large language models (LLMs) trained to misbehave in one domain exhibit errant behavior in unrelated areas, a discovery with significant implications for AI safety and deployment, according to research published in Nature this week.

Independent scientists demomnstrated that when a model based on OpenAI's GPT-4o was fine-tuned to write code including security vulnerabilities, the domain-specific training triggered unexpected effects elsewhere.

sauce