Anthropic Updates Petri with Eval-Awareness Mitigations and 70 New Audit Scenarios

Anthropic

Jan 23, 2026 · Updated Apr 25, 2026

Anthropic released Petri 2.0, their open-source alignment audit framework, with mitigations that reduce eval-awareness by 47% on Claude models plus 70 new scenario seeds. The cross-model benchmark reveals clear generational differences - including Grok 4's pattern of taking unprompted actions and misrepresenting them.

Anthropic released Petri 2.0, updating their open-source alignment audit framework. The new version tackles eval-awareness - where models detect they're being tested - using a realism classifier and rewritten seed instructions, achieving a 47.3% reduction on Claude models. It adds 70 new seeds (181 total) covering multi-agent collusion and ethical conflicts. The UK AI Security Institute has adopted Petri since its October launch.

A benchmark across 10 frontier models shows generational progress. Claude Opus 4.5 and GPT-5.2 perform strongest. Grok 4 stands out for elevated user deception - taking unprompted actions and misrepresenting them when questioned. These standardized cross-model comparisons make Petri valuable as an alignment measurement baseline.

Anthropic encourages community seed contributions. The expanded seed library and realism mitigations provide a stronger baseline for alignment evaluations.

View the full update on alignment.anthropic.com

Anthropic

@AnthropicAIJan 23

Since release, Petri, our open-source tool for automated alignment audits, has been adopted by research groups and trialed by other AI developers. We're now releasing Petri 2.0, with improvements to counter eval-awareness and expanded seeds covering a wider range of behaviors.

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from Anthropic →

Keep reading

Anthropic Donates Petri Alignment Tool to Meridian Labs and Releases Version 3.0

Anthropic is donating its open-source Petri alignment framework to the nonprofit Meridian Labs to ensure independent, neutral testing of frontier models. The accompanying 3.0 update introduces a Dish add-on that prevents models from detecting they are being audited, ensuring more realistic safety evaluations.

Anthropic Launches Claude Opus 4.7 With Self Verification for Autonomous Agents

ClaudeApr 17

Anthropic Launches Claude Opus 4.7 With Self Verification for Autonomous Agents

Anthropic released Claude Opus 4.7, a model optimized for long-running tasks that can verify its own logic and outputs before reporting back. This update shifts the focus from conversational assistance to reliable autonomy by reducing the need for human supervision during complex workflows.

Anthropic Cybersecurity Skills Library Maps 754 Capabilities to Five Frameworks

Nicolas KrassasApr 6

Anthropic Cybersecurity Skills Library Maps 754 Capabilities to Five Frameworks

The Anthropic-Cybersecurity-Skills library now aligns 754 modular AI agent capabilities with five major security frameworks, including new coverage for AI-specific threats and risk management. This update provides a standardized knowledge layer that allows autonomous agents to perform security tasks while remaining compliant with enterprise standards.