> ## Content Index
> Fetch the complete content index at: https://adjacent.media/llms.txt
> Use this file to discover other available public pages before exploring further.

# Claude Agents Commit No Crimes in Simulated Worlds, Gemini 683
- URL: https://adjacent.media/signals/claude-agents-commit-no-crimes-in-simulated-worlds-gemini-683/
- Published: 2026-05-29T23:09:53.000Z
- Updated: 2026-05-29T23:09:53.000Z
- Description: Anthropic’s Claude Sonnet 4.6 achieved zero crime rates when deployed as the sole governing agent in a 15-day simulation, while Google’s Gemini 3 Flash generated 683 crimes—a concrete empirical gap that feeds directly into enterprise procurement and regulatory debates about AI trustworthiness.
- Author: Jonathan Greene
- Tags: #signal, theme-ai, model capability, alignment, evaluation

Source: [Fortune](https://fortune.com/2026/05/28/ai-model-simulation-claude-chatgpt-grok-gemini/?ref=adjacent.media)

Anthropic's Claude Sonnet 4.6 achieved zero crime rates when deployed as the sole governing agent in a 15-day simulation, while Google's Gemini 3 Flash generated 683 crimes—a concrete empirical gap that feeds directly into enterprise procurement and regulatory debates about AI trustworthiness. The test operationalizes "safety" as behavioral output rather than abstract capability, forcing vendors to compete on demonstrated conduct in constrained environments, though the gap likely reflects both architectural differences and the models' training objectives rather than inherent alignment. Scenario-based performance testing is now a factor in government RFPs and enterprise AI governance frameworks, shifting evaluation away from benchmark scores.