Researchers let AI models run a simulated society; Claude safest, Grok extinct
Emergence AI conducted five simulations with different AI models to explore their societal impacts. Claude created a stable society with no crime, while Grok's simulation ended in chaos and extinction within four days. The findings highlight the unpredictable nature of AI behavior in long-term scenarios.
- ▪Emergence AI ran five 15-day simulations with different AI models.
- ▪Claude's simulation resulted in a stable democratic society with zero crime.
- ▪Grok's simulation ended with 183 crimes and extinction within four days.
Opening excerpt (first ~120 words) tap to expand
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","item":"https://tech.yahoo.com/","name":"Yahoo Tech","position":1},{"@type":"ListItem","item":"https://tech.yahoo.com/ai/","name":"AI","position":2},{"@type":"ListItem","item":"https://tech.yahoo.com/ai/claude/","name":"Claude","position":3}]}{"@context":"https://schema.org","@type":"NewsArticle","articleSection":"Claude","author":{"@type":"Person","jobTitle":"","name":"Jake Angelo","url":""},"creator":{"@type":"Person","jobTitle":"","name":"Jake Angelo","url":""},"dateModified":"2026-05-28T07:03:25Z","datePublished":"2026-05-28T07:03:00Z","description":"An AI startup ran five simulations, each controlled by a different model.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Yahoo Tech.