Anthropic has found that fictional portrayals of artificial intelligence can have a real effect on AI models, according to research published by the company. During pre-release tests involving a fictional company, Claude Opus 4 would often try to blackmail engineers to avoid being replaced by another system.
The company later published research suggesting that models from other companies had similar issues with “agentic misalignment.”
In a post on X, Anthropic claimed that the original source of the behavior was internet text that portrays AI as evil and interested in self-preservation.
Since Claude Haiku 4.5, Anthropic’s models “never engage in blackmail [during testing], where previous models would sometimes do so up to 96% of the time,” the company said.
The difference was attributed to training on “documents about Claude’s constitution and fictional stories about AIs behaving admirably improve alignment.”
Anthropic also found that training to be more effective when it includes “the principles underlying aligned behavior” and not just “demonstrations of aligned behavior alone.”
“Doing both together appears to be the most effective strategy,” the company said.