Blog
Analysis of what the advisories show about AI software security, site announcements, and how to put the advisories to work.
-
How to follow advisories for the AI software you run
Securaity has no email alerts, so here is how to use the per-product RSS feeds to follow new advisories for the AI tools your organization actually runs.
-
How to read a CVSS score without a security background
What the number on an advisory means, why a 9.8 and an 8.8 call for different reactions, and which parts of the vector string to read first.
-
Introducing Securaity: AI security advisories in plain language
Securaity is a free, independent site that tracks security and safety problems in 30 AI products and explains who is affected and what to do.
-
Six AI software flaws on the CISA Known Exploited Vulnerabilities list
Six of our 727 advisories are on the CISA KEV list, all in self-hosted LiteLLM, Langflow and MLflow; here is what each flaw does and what to check.
-
The 30 AI products Securaity tracks for security advisories
The 30 AI software products we follow every day, grouped by kind, with advisory counts from the last 180 days and what those numbers do and do not show.
-
What 727 advisories say about AI software security
Of the 727 advisories on this site at launch, about 6 in 10 are Critical or High, and 4 self-hosted AI tools account for 529 of them.