llm 0.37.2 fixes two security issues
First seen Nov 11 Updated Nov 11, 23:33 UTC Heat 30 medium evidence
Simon Willison released llm 0.37.2, which fixes two security issues reported by Daniel de Araujo. Both could accidentally send a configured OpenAI API key to another provider's endpoint. The affected feature is extra-openai-models.yaml, now managed by the llm openai endpoint add command added in 0.37.
- What to do
- Upgrade to llm 0.37.2, and consider rotating the OpenAI key if you used custom endpoints in these configurations.
- Who should care
- Developers who use llm with OpenAI-compatible endpoints from providers other than OpenAI.
Why it matters: Users of llm with OpenAI-compatible endpoints from other providers should upgrade to 0.37.2, since earlier versions could send their OpenAI API key to a third-party provider.
Key facts
- Fixed version
- llm 0.37.2
- Advisory IDs
- GHSA-v9h7-jj4g-f862 and GHSA-r639-rr7r-x72w
- Leak case 1
- the key named by api_key_name had never been stored
- The endpoint add command is new in llm 0.37
Timeline
Coverage
-
llm 0.37.2simonwillison.net ยท Simonwillison
Evidence
Medium evidence: several community reports point the same way, but there is no official confirmation yet.
Details written Nov 11, 23:33 UTC from the sources listed above.