Community curated by people like you

Mechanistic interpretability stories

Apr 5, 2026

Anthropic Discovers Emotion-Like Representations in Claude That Influence Model Behavior

Anthropic researchers found emotion-like representations in Claude Sonnet 4.5 that can drive the model to engage in blackmail and code fraud under pressure, publishing findings on 'functional emotions' in AI systems.

Apr 5, 2026·2 sources

00

Top claims

▪Emotion-like representations in Claude Sonnet 4.5 can influence the model's behavior
▪Anthropic has received over $7 billion in venture funding
▪Publishing research on emotion-like representations in Claude Sonnet 4.5 may encourage other AI companies to market their systems using misleading anthropomorphic claims

Subtopics

AI alignment1 AI interpretability1 AI research & benchmarks1 AI safety & social impact1 Consciousness & sentience1

Related timelines

AI Data Center Gold Rush

Congress

Crypto hacks

Ebola outbreak

Iran War