I’m excited to announce the first significant update to the OWASP Top 10 for Large Language Model Applications! The field of LLMs is moving fast and this version (v1.1) includes numerous enhancements and updates from the last two months. The improvements certainly include grammar and readability, but there’s much more. In this version, we’ve updated examples, streamlined the risk definitions, and clarified the differences between related risk types.

Thanks to Ads Dawson from Cohere for leading the effort for this update. Overall, 35 enhancement suggestions were raised by the community and addressed. This version is also our first to include a visual diagram showing the risks of a typical LLM app architecture.

Want to stay up to speed on the latest? You can grab the new version of The List documents from our Project Site. You can keep track of our progress on LinkedIn at OWASP Top 10 For Large Language Model Applications or join our mailing list if that’s your preference.

Linkedin post link:
New Release of OWASP Top 10 for LLM Apps