CVE-2026-31254
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04)...
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains a code injection vulnerability (CWE-94) in its training script. The script registers the Python eval() function as a Hydra configuration resolver under the name eval. This allows configuration files to execute arbitrary Python code via the ${eval:...} syntax. An attacker can exploit this by providing a malicious configuration file, leading to arbitrary code execution when the training script is run with that configuration.
Attack profile
How easy it is to exploit
- Attack vector
- Network
- Complexity
- Low
- Privileges needed
- None
- User interaction
- None
What an attacker can affect
- Confidentiality
- Low
- Integrity
- Low
- Availability
- Low
Exploit likelihood
What each source reports
CVE List Score used
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains a code injection vulnerability (CWE-94) in its training script. The script registers the Python eval() function as a Hydra configuration resolver under the name eval. This allows configuration files to execute arbitrary Python code via the ${eval:...} syntax. An attacker can exploit this by providing a malicious configuration file, leading to arbitrary code execution when the training script is run with that configuration.
- Record
CVE-2026-31254- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L - Weakness
- CWE-95
- Updated
- 2026-11-02
EUVDEUVD-2026-29101
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains a code injection vulnerability (CWE-94) in its training script. The script registers the Python eval() function as a Hydra configuration resolver under the name eval. This allows configuration files to execute arbitrary Python code via the ${eval:...} syntax. An attacker can exploit this by providing a malicious configuration file, leading to arbitrary code execution when the training script is run with that configuration.
- Record
EUVD-2026-29101- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L - Updated
- 2026-11-02
GitHub AdvisoriesGHSA-9hgr-qwgw-x8pp
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04)...
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains a code injection vulnerability (CWE-94) in its training script. The script registers the Python eval() function as a Hydra configuration resolver under the name eval. This allows configuration files to execute arbitrary Python code via the ${eval:...} syntax. An attacker can exploit this by providing a malicious configuration file, leading to arbitrary code execution when the training script is run with that configuration.
- Record
GHSA-9hgr-qwgw-x8pp- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L - Weakness
- CWE-95
- Updated
- 2026-11-02
NVD
The flash-attention project thru commit e724e2588cbe754beb97cf7c011b5e7e34119e62 (2025-13-04) contains a code injection vulnerability (CWE-94) in its training script. The script registers the Python eval() function as a Hydra configuration resolver under the name eval. This allows configuration files to execute arbitrary Python code via the ${eval:...} syntax. An attacker can exploit this by providing a malicious configuration file, leading to arbitrary code execution when the training script is run with that configuration.
- Record
CVE-2026-31254- CVSS 3.1
- 7.3 High
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:L - Weakness
- CWE-95
- Updated
- 2026-11-02
History
| Recorded | Changed | Sources |
|---|---|---|
| first indexed | CVE List, EUVD, GitHub Advisories, NVD |
Dates are when Every CVE recorded each change. Upstream dates may differ.
In short
CVE-2026-31254 is a high vulnerability (CVSS 7.3), published May 11, 2026. No source lists public exploit code. EPSS puts the chance of exploitation in the next 30 days at 0.25%, the 16th percentile. No source names a fixed version.