CVE-2025-12695
Insecure configuration in DSPy lead to arbitrary file read when running untrusted code inside the sandbox
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
Attack profile
How easy it is to exploit
- Attack vector
- Network
- Complexity
- High
- Privileges needed
- None
- User interaction
- None
What an attacker can affect
- Confidentiality
- High
- Integrity
- None
- Availability
- None
Exploit likelihood
What each source reports
CVE List Score used
Insecure configuration in DSPy lead to arbitrary file read when running untrusted code inside the sandbox
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
- Record
CVE-2025-12695- CVSS 3.1
- 5.9 Medium
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N - Weakness
- CWE-653
- Updated
- 2026-11-02
EUVDEUVD-2025-37741
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
- Record
EUVD-2025-37741- CVSS 3.1
- 5.9 Medium
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N - Updated
- 2026-11-02
GitHub AdvisoriesGHSA-vvw2-h478-xwr3
DSPy does not properly restrict file reads
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
- Record
GHSA-vvw2-h478-xwr3- CVSS 3.1
- 5.9 Medium
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N - Weakness
- CWE-653
- Updated
- 2026-11-02
NVD
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
- Record
CVE-2025-12695- CVSS 3.1
- 5.9 Medium
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N - Weakness
- CWE-653
- Updated
- 2026-11-05
OSVGHSA-vvw2-h478-xwr3
DSPy does not properly restrict file reads
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
- Record
GHSA-vvw2-h478-xwr3- CVSS 3.1
- 5.9 Medium
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N - Weakness
- CWE-653
- Updated
- 2026-11-02
OSVPYSEC-2026-1318
DSPy does not properly restrict file reads
The overly permissive sandbox configuration in DSPy allows attackers to steal sensitive files in cases when users build an AI agent which consumes user input and uses the “PythonInterpreter” class.
- Record
PYSEC-2026-1318- CVSS 3.1
- 5.9 Medium
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N - Updated
- 2026-11-02
Affected packages
| Package | Affected | Fixed |
|---|---|---|
dspyPyPI |
through 3.0.3 | None known |
dspyPyPI |
through 3.0.3, plus listed versions 0.0.10.0.20.0.30.1.30.1.4 Show 45 moreShow less0.1.52.5.12.5.102.5.112.5.122.5.132.5.142.5.152.5.162.5.172.5.182.5.192.5.22.5.202.5.212.5.222.5.232.5.242.5.252.5.262.5.272.5.282.5.292.5.32.5.302.5.312.5.322.5.332.5.342.5.352.5.362.5.372.5.382.5.392.5.42.5.402.5.412.5.422.5.432.5.52.5.62.5.72.5.82.5.92.6.0 |
None known |
History
| Recorded | Changed | Sources |
|---|---|---|
| facts | NVD | |
| first indexed | CVE List, EUVD, GitHub Advisories, NVD, OSV |
Dates are when Every CVE recorded each change. Upstream dates may differ.
In short
CVE-2025-12695 is a medium vulnerability (CVSS 5.9) in dspy, published November 4, 2025. No source lists public exploit code. EPSS puts the chance of exploitation in the next 30 days at 0.34%, the 27th percentile. No source names a fixed version.
References
Left out here: 45 versions. The JSON record has them.