The page opens by saying data privacy and security are LiteLLM's top priority
The page opens with a general statement that LiteLLM treats the privacy and security of your data as its top priority and handles the data you share with it with the highest level of diligence. The statement names no specific measure, data type or user group. The concrete statements come in the sections that follow, and several of those sections address self-hosted users only.
Supporting quotations
safeguarding your data privacy and security is our top priority
handle it with the highest level of diligence
Not stated here
The opening sentence refers to "the data you share with us" but does not say what data that is or which users share it. The sections that follow say a self-hosting user sends no data to LiteLLM servers except through telemetry that the user points there.
Security & Safeguards · Procedure
The page gives support@berri.ai for security inquiries and krrish@berri.ai for incidents and questions
The page tells readers to contact support@berri.ai for security inquiries. The vendor information section separately gives krrish@berri.ai as the point of contact for security incidents and for general security-related questions. Vulnerability reports have their own route, described in the vulnerability reporting card.
Supporting quotations
For security inquiries, please contact us at
Point of contact email address for security incidents:
Point of contact email address for general security-related questions:
Qualifications & exceptions
The support@berri.ai address appears in the self-hosted security measures section; the krrish@berri.ai addresses appear in the vendor information list.
Not stated here
The page does not say how the two addresses differ or which one to use for a given kind of question.
Security & Safeguards · Procedure
Vulnerability reports go preferably to GitHub Security Advisories, with email as a slower fallback
The page says the canonical vulnerability disclosure policy is the security.md file in the LiteLLM repository, which includes information about the bug bounty. It says the preferred way to report a vulnerability is a private report through GitHub Security Advisories. If you cannot use GitHub, emailing support@berri.ai with the same details works, but may be triaged more slowly. In the vendor questions, the page answers Yes to whether a vulnerability disclosure policy is in place.
Supporting quotations
The canonical vulnerability disclosure policy lives in
which includes information about our bug bounty
The preferred channel for reporting a vulnerability is to file a private vulnerability report through
If you cannot use GitHub, emailing support@berri.ai with the same details works in a pinch, though may be triaged more slowly.
Does the vendor have a vulnerability disclosure policy in place?
Qualifications & exceptions
The security.md file is named as the canonical policy; this page summarizes it only in part.
Email is a fallback for people who cannot use GitHub.
Not stated here
The security.md file is linked but not supplied here, so the bug bounty terms and any differences from this page are unknown.
Security & Safeguards · Restriction or exception
All vulnerability reports must include a video or screen recording of the full exploit
The page says that, to combat low-quality AI spam, all vulnerability reports require a video or screen recording that demonstrates the full exploit. The wording says all reports, so it covers the GitHub route and the email fallback described on the same page.
Supporting quotations
in order to combat low-quality AI spam, all vulnerability reports require a video or screen recording of you demonstrating the full exploit
Qualifications & exceptions
Stated as a requirement for all vulnerability reports, with the stated aim of combating low-quality AI spam.
Not stated here
The page does not say what happens to a report that arrives without a video or screen recording. The linked security.md policy is not supplied and could add or change requirements.
Security & Safeguards · Company practice
LiteLLM runs grype security scans on all built Docker images
The page says LiteLLM runs grype security scans on all built Docker images and shows a current status line of passing with no high or critical severity vulnerabilities found. In the vendor questions it answers Yes to performing vulnerability scans and says regular scans are conducted as detailed in the Vulnerability Scanning section.
Supporting quotations
LiteLLM runs grype security scans on all built Docker images.
Current Status: ✅ Passing. 0 High/Critical severity vulnerabilities found.
Does the vendor perform vulnerability scans?
Yes, regular vulnerability scans are conducted as detailed in the Vulnerability Scanning section.
Qualifications & exceptions
The scanning described is for built Docker images.
The status line is a point-in-time statement.
Not stated here
The status line has no date, so the time it describes is unknown. "Regular" is not defined, and the page does not state a scan frequency or say whether anything other than Docker images is scanned. The CI/CD link the page cites is not supplied.
Security & Safeguards · Company practice
LiteLLM says it holds SOC 2 Type II and makes its report available through its Trust Center
In answer to whether the vendor has been audited or certified, the page says SOC 2 Type II and that the current report is available through the LiteLLM Trust Center.
Supporting quotations
Has the Vendor been audited / certified?
SOC 2 Type II. Our current report is available through the LiteLLM Trust Center.
Qualifications & exceptions
The report is reached through the LiteLLM Trust Center, not on this page.
Not stated here
The page does not say which product, entity or period the report covers, or what conditions apply to getting the report. The Trust Center is linked but not supplied.
Security & Safeguards · Company practice
LiteLLM says it has an information security management system and uses CodeQL
In answer to whether an information security management system has been implemented, the page says Yes, with CodeQL and an ISMS covering multiple security domains.
Supporting quotations
Has an information security management system been implemented?
and an ISMS covering multiple security domains
Not stated here
The page does not list the security domains the ISMS covers or name a standard it follows.
Security & Safeguards · Company practice
LiteLLM says it logs key events such as authentication, creation and updates through audit logs
In answer to whether logging of key events (authentication, creation and update changes) occurs, the page says Yes and points to audit logs documentation. The page's description of LiteLLM Enterprise also lists audit logs among its features.
Supporting quotations
Is logging of key events - auth, creation, update changes occurring?
Yes - we have
audit logs
Qualifications & exceptions
The question concerns key events: authentication, creation and update changes.
Not stated here
The page does not say whether audit logs are available in the free self-hosted software or only in LiteLLM Enterprise, or whether they are on by default. The documentation link, which refers to switching audit logs on, is not supplied.
LiteLLM says it has an incident management program and shares its Incident Response Policy on request
In answer to whether it has an established cybersecurity incident management program, the page says Yes, and that its Incident Response Policy is available upon request.
Supporting quotations
Does the Vendor have an established Cybersecurity incident management program?
Yes, Incident Response Policy available upon request.
Qualifications & exceptions
The policy is provided only on request.
Not stated here
The page does not say who may request the policy, how to request it, or what it contains. The policy is not supplied.