CVE-2026-72742

HIGHCVSS 8.6/10EPSS 0.51%

Last modified

CVE-2026-72742 is a high-severity vulnerability rated 8.6/10 on the CVSS scale. DSPy 3.3.0b1 contains a file exfiltration vulnerability in the Image and Audio output field adapters that allows attackers with influence over language model outputs to read arbitrary local files by injecting a filesystem path into the url field of a parsed Image or Audio typed output. The JSONAdapter and ChatAdapter parse untrusted language model completions through parse_value into TypeAdapter validation, which triggers encode_image or encode_audio to read and base64-encode any local file path via the os.path.isfile branch in image.py and audio.py, subsequently embedding the file contents into outgoing prompt messages sent to the attacker-controlled model endpoint.. EPSS estimates a 0.51% chance of exploitation in the next 30 days.

Description

DSPy 3.3.0b1 contains a file exfiltration vulnerability in the Image and Audio output field adapters that allows attackers with influence over language model outputs to read arbitrary local files by injecting a filesystem path into the url field of a parsed Image or Audio typed output. The JSONAdapter and ChatAdapter parse untrusted language model completions through parse_value into TypeAdapter validation, which triggers encode_image or encode_audio to read and base64-encode any local file path via the os.path.isfile branch in image.py and audio.py, subsequently embedding the file contents into outgoing prompt messages sent to the attacker-controlled model endpoint.

Metrics

Weakness Enumeration

Affected Software

Source: CNA advisory (CVE.org). NVD analysis pending.

VendorProductVersions
Stanford NLPDSPy<= 3.3.0b1

References

Timeline

Published
Last Modified
Status
Awaiting Analysis

Frequently Asked Questions

What is CVE-2026-72742?
DSPy 3.3.0b1 contains a file exfiltration vulnerability in the Image and Audio output field adapters that allows attackers with influence over language model outputs to read arbitrary local files by injecting a filesystem path into the url field of a parsed Image or Audio typed output. The JSONAdapter and ChatAdapter parse untrusted language model completions through parse_value into TypeAdapter validation, which triggers encode_image or encode_audio to read and base64-encode any local file path via the os.path.isfile branch in image.py and audio.py, subsequently embedding the file contents into outgoing prompt messages sent to the attacker-controlled model endpoint.
How severe is CVE-2026-72742?
CVE-2026-72742 has a CVSS score of 8.6/10 (HIGH severity). The EPSS model estimates a 0.51% probability of exploitation in the next 30 days.
How do I fix CVE-2026-72742?
Check the vendor references and advisories linked above for patched versions and mitigation guidance. You can also run a Strix scan to test if your systems are affected.

How Strix Helps

Related CVEs from 2026

Are you affected by CVE-2026-72742?

Run a free Strix scan to check your systems for this vulnerability.

Scan your code now

Source: NVD / NIST