Disclosed weakness · no CVE recorded
LiteLLM response comparisons could bypass user and team ownership checks
LiteLLM · LiteLLM proxy · Responses API
First documented disclosure · Review checkpoint
Summary
A caller could submit a provider response identifier for cache comparison without the ownership check applied to normal response references.
Severity and exploitation evidence
No CVSS score recorded in this review.
Exploitation not established: The reviewed sources describe a correction or controlled reproduction; malicious exploitation against users is not established.
Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.
How the weakness works
Comparison identifiers were forwarded without the proxy ownership validation; a raw provider identifier could be accepted across users or teams.
Conditions required
An authenticated caller knows a provider response identifier and reaches the comparison API on an affected proxy configuration.
Potential impact
A caller could obtain comparison diagnostics against a response outside its user or team scope. This is not evidence of full response-content disclosure.
Affected versions and configurations
The reviewed upstream report does not establish a complete affected-release range.
Fix and mitigation status
1.110.0 contains the correction; the earliest corrected release on other lines is unverified.
Practical next steps
Upgrade to 1.110.0 and leave unmanaged response identifiers disabled unless the operator deliberately accepts their weaker ownership guarantees.
Operating lesson
Apply ownership checks to every field that refers to an earlier response.
Evidence and limits
The October 28 public PR precedes the October 29 issue and October 31 replacement PR. A controlled provider-backed test returned cache diagnostics for raw cross-user references before correction. Tagged verification: comparison references are passed through authorize_response_id before provider dispatch. Existing explicit security bypass modes remain documented limitations.
Sources
- fix(responses): send OpenAI the original response id in prompt_cache_options.comparison_response_id
Primary source ·
- [Bug]: Responses API rejects a LiteLLM response id in prompt_cache_options.comparison_response_id
Primary source ·
- fix(responses)!: authorize and decode comparison response IDs
Primary source ·
- v1.110.0 release notes
Primary source ·
- LiteLLM 1.110.0 comparison ownership checks
Primary source ·