According to Anthropic (@AnthropicAI), Petri, their open-source platform for automated AI alignment audits, has seen significant adoption by research groups and AI developers since its initial release. The newly launched Petri 2.0 introduces key improvements such as enhanced countermeasures against eval-awareness—where AI systems may adapt behavior during evaluation—and expands its seed set to audit a broader spectrum of AI behaviors. These updates are designed to streamline large-scale, automated safety assessments, providing AI researchers and businesses with a more reliable method for detecting misalignment in advanced models. Petri 2.0 aims to support organizations in proactively identifying risks and ensuring responsible AI deployment, addressing growing industry demands for robust AI safety tools (source: AnthropicAI on Twitter, January 23, 2026).