CVE-2024-7760 is a critical (CVSS 9.6) cross-site request forgery flaw in the tracking server of aimhubio/aim 3.22.0, an experiment tracker for machine learning. Overly permissive CORS settings allow requests from any origin. The record names no fixed release, so limit who can reach the server and check the project's releases.
What's affected
The record names aimhubio/aim version 3.22.0 and its tracking server. The cause is a CORS configuration that allows cross-origin requests from all origins, which makes cross-site request forgery possible against every endpoint of the server.
The description says this can be chained with other weaknesses in the same server, including remote code execution, denial of service and arbitrary file read and write. That chaining is why the score is critical even though the request forgery itself needs a user to visit a hostile page (the vector marks user interaction as required).
The only source in the record is a huntr bounty report. We could not load it in an automated fetch, so a fixed version is not confirmed here. Read the Aim release notes and the report before relying on any specific version number.
How to check if you're exposed
Check which version of the package your environments have installed.
pip show aim
pip list | grep -i aim
Run the commands inside each virtual environment, container image and notebook environment that might use it, because the system Python often differs from the project one. For a lockfile or requirements file, search for aim and read the pinned version.
Also check whether anyone runs the Aim server, which is started with the aim server command, and where it listens. A tracking server bound to a laptop or a shared host is the target here.
Safeguard's free in-browser dependency scan at gold.safeguard.sh/scan can read a dependency manifest and flag known vulnerable versions without installing anything.
What to do now
- List every environment that has Aim installed and note the version.
- Find any running Aim tracking server and who can reach it.
- Do not expose the tracking server to the internet; put it behind a VPN or an authenticating reverse proxy.
- Upgrade to a release that the project says fixes the CORS issue, after confirming in its release notes.
- Ask users not to browse untrusted sites in a browser that can reach the internal tracking server.
Why it matters
Browsers let any web page send requests to any address, including internal ones. CORS rules exist so that a server can say which origins may read responses and send credentialed requests. A server that answers yes to every origin gives up that protection.
For internal ML tools this matters because they often run on trusted networks without a login. A page visited by an engineer can then act on the tracking server from the engineer's own browser. The weakness class is cross-site request forgery. The fix is to allow only named origins and to require a token on state changing calls.
The weakness class is CWE-352: Cross-Site Request Forgery. The CVSS vector in the record is CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:H, which scores 9.6. This CVE is not listed in the CISA Known Exploited Vulnerabilities catalog at the time of writing.
Try it yourself
Click through it at your own pace, or open it full screen.
Quick check
Does an attacker need network access to the server?
They need a victim whose browser can reach it. The vector marks user interaction as required.
Is a fixed version confirmed?
Not from the sources checked here. Read the project release notes.
Sources
Last reviewed: 19 October 2026.
