Azure/PyRIT
PyRIT (Python Risk Identification Tool) is an open-source framework from Microsoft/Azure that lets security professionals automate red-teaming of generative AI systems, generating adversarial prompts, scoring model responses, and probing for harms like jailbreaks, harmful content generation, and other LLM risks. It provides a modular pipeline (orchestrators, targets, converters, scorers) to systematically test chatbots and other GenAI applications for vulnerabilities.
What it is
PyRIT (Python Risk Identification Tool for generative AI), the Azure open-source red-teaming framework for GenAI systems, was archived by its owner on 27 March 2026. GitHub now lists the repository as read-only.
Preconditions / Affected
The archive status matters to anyone planning to adopt PyRIT or already depending on it. A read-only repository accepts no new commits, issues or pull requests. The page shows no sign of fixes, new attack converters or scorers, or compatibility updates after that date.
What to do
Check the archive status before building a red-teaming workflow on PyRIT. If you already use it, pin the version you have tested and plan for maintaining it yourself or moving to another tool. The page names no successor, fork or migration path, so look for one outside this repository.
Caveats
The fetched page gives no reason for the archiving. It does not say whether the project moved elsewhere, and it does not say whether the last release is stable or has known gaps. The page text also contains no install commands, version numbers or feature documentation beyond the one-line project description.
Related in AI Security
