Hi openeuler-ci-bot, welcome to the openEuler Community.
I'm the Bot here serving you. You can find the instructions on how to interact with me at Here.
If you have any questions, please contact the SIG: A-Tune, and any of the maintainers: @Vchanger , @znzjugod , @pshysimon , @chendh6 , @hubin95 , @yangyongguang


@hubin95 ,@yangyongguang ,@Vchanger ,@znzjugod ,@chendh6 ,@pshysimon
issue处理注意事项:
1. 当前issue受影响的分支提交pr时, 须在pr描述中填写当前issue编号进行关联, 否则无法关闭当前issue;
2. 模板内容需要填写完整, 无论是受影响或者不受影响都需要填写完整内容,未引入的分支不需要填写, 否则无法关闭当前issue;
3. 以下为模板中需要填写完整的内容, 请复制到评论区回复, 注: 内容的标题名称(影响性分析说明, openEuler评分, 受影响版本排查(受影响/不受影响), 修复是否涉及abi变化(是/否), 原因说明)不能省略,省略后cve-manager将无法正常解析填写内容.
影响性分析说明:
openEuler评分: (评分和向量)
受影响版本排查(受影响/不受影响):
1.master(2.12.0):
2.openEuler-20.03-LTS-SP4:
3.openEuler-22.03-LTS-SP3(2.10.0):
4.openEuler-22.03-LTS-SP4(2.10.0):
5.openEuler-24.03-LTS(2.12.0):
6.openEuler-24.03-LTS-Next(2.12.0):
7.openEuler-24.03-LTS-SP1(2.12.0):
8.openEuler-24.03-LTS-SP2(2.12.0):
修复是否涉及abi变化(是/否):
1.master(2.12.0):
2.openEuler-20.03-LTS-SP4:
3.openEuler-22.03-LTS-SP3(2.10.0):
4.openEuler-22.03-LTS-SP4(2.10.0):
5.openEuler-24.03-LTS(2.12.0):
6.openEuler-24.03-LTS-Next(2.12.0):
7.openEuler-24.03-LTS-SP1(2.12.0):
8.openEuler-24.03-LTS-SP2(2.12.0):
原因说明:
1.master(2.12.0):
2.openEuler-20.03-LTS-SP4:
3.openEuler-22.03-LTS-SP3(2.10.0):
4.openEuler-22.03-LTS-SP4(2.10.0):
5.openEuler-24.03-LTS(2.12.0):
6.openEuler-24.03-LTS-Next(2.12.0):
7.openEuler-24.03-LTS-SP1(2.12.0):
8.openEuler-24.03-LTS-SP2(2.12.0):
原因说明填写请参考下方表格(注意:版本是否受影响和版本的原因说明必须对应,例如master版本分支受影响,那原因说明只能是受影响对应的原因之一!):
| 分支状态 | 原因说明 |
|---|---|
| 受影响 | 正常修复 |
| 受影响 | 暂不修复-漏洞仍在分析中 |
| 受影响 | 暂不修复-暂无解决方案或补丁 |
| 受影响 | 暂不修复-待升级版本修复 |
| 受影响 | 不修复-超出修复范围 |
| 受影响 | 不修复-特殊原因导致不再修复 |
| 不受影响 | 不受影响-组件不存在 |
| 不受影响 | 不受影响-已有内置的内联控制或缓解措施 |
| 不受影响 | 不受影响-漏洞代码不能被攻击者触发 |
| 不受影响 | 不受影响-漏洞代码不在执行路径 |
| 不受影响 | 不受影响-漏洞代码不存在 |
issue处理具体操作请参考:
https://gitee.com/openeuler/cve-manager/blob/master/cve-vulner-manager/doc/md/manual.md
pr关联issue具体操作请参考:
https://gitee.com/help/articles/4142


| 参考网址 | 关联pr | 状态 | 补丁链接 |
|---|---|---|---|
| https://nvd.nist.gov/vuln/detail/CVE-2025-9906 | |||
| https://ubuntu.com/security/CVE-2025-9906 | |||
| https://www.opencve.io/cve/CVE-2025-9906 | |||
| https://bugzilla.redhat.com/show_bug.cgi?id=CVE-2025-9906 | |||
| https://security-tracker.debian.org/tracker/CVE-2025-9906 | |||
| http://www.cnnvd.org.cn/web/vulnerability/queryLds.tag?qcvCnnvdid=CVE-2025-9906 |
说明:抱歉,当前工具暂未找到推荐补丁,请人工查找或者之后评论'/find-patch'尝试再次查找。
若人工查找到补丁,烦请在此issue下评论 '/report-patch 参考网址 补丁链接1,补丁链接2' 便于我们不断优化工具,不胜感激。
如 /report-patch https://security-tracker.debian.org/tracker/CVE-2021-3997 https://github.com/systemd/systemd/commit/5b1cf7a9be37e20133c0208005274ce4a5b5c6a1


@ 经过 cve-manager 解析, 已分析的内容如下表所示:
| 状态 | 分析项目 | 内容 |
|---|---|---|
| 已分析 | 1.影响性分析说明 | The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=True. One can create a specially crafted .keras model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special config.json (a file within the .keras archive) that will invoke keras.config.enable_unsafe_deserialization() to disable safe mode. Once safe mode is disable, one can use the Lambda layer feature of keras, which allows arbitrary Python code in the form of pickled code. Both can appear in the same archive. Simply the keras.config.enable_unsafe_deserialization() needs to appear first in the archive and the Lambda with arbitrary code needs to be second. |
| 已分析 | 2.openEulerScore | 7.3 |
| 已分析 | 3.openEulerVector | AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:H/A:H |
| 已分析 | 4.受影响版本排查 | master:受影响,openEuler-20.03-LTS-SP4:受影响,openEuler-22.03-LTS-SP3:受影响,openEuler-22.03-LTS-SP4:受影响,openEuler-24.03-LTS:受影响,openEuler-24.03-LTS-Next:受影响,openEuler-24.03-LTS-SP1:受影响,openEuler-24.03-LTS-SP2:受影响,openEuler-24.03-LTS-SP3:受影响 |
| 已分析 | 5.是否涉及abi变化 | master:否,openEuler-20.03-LTS-SP4:否,openEuler-22.03-LTS-SP3:否,openEuler-22.03-LTS-SP4:否,openEuler-24.03-LTS:否,openEuler-24.03-LTS-Next:否,openEuler-24.03-LTS-SP1:否,openEuler-24.03-LTS-SP2:否,openEuler-24.03-LTS-SP3:否 |
| 已分析 | 6.原因说明 | master:暂不修复-暂无解决方案或补丁,openEuler-20.03-LTS-SP4:暂不修复-暂无解决方案或补丁,openEuler-22.03-LTS-SP3:暂不修复-暂无解决方案或补丁,openEuler-22.03-LTS-SP4:暂不修复-暂无解决方案或补丁,openEuler-24.03-LTS:暂不修复-暂无解决方案或补丁,openEuler-24.03-LTS-Next:暂不修复-暂无解决方案或补丁,openEuler-24.03-LTS-SP1:暂不修复-暂无解决方案或补丁,openEuler-24.03-LTS-SP2:暂不修复-暂无解决方案或补丁,openEuler-24.03-LTS-SP3:暂不修复-暂无解决方案或补丁 |
请确认分析内容的准确性, 确认无误后, 您可以进行后续步骤, 否则您可以继续分析.


🔍 CVE 项目匹配性分析结果
CVE: CVE-2025-9906
针对分支: master
结果: ❌ 不匹配
原因: CVE-2025-9906 仅影响 Keras 3.x 版本(3.0.0 到 3.10.0),项目当前版本 2.12.0 属于 Keras 2.x 系列,不在受影响范围内。OSV 数据明确列出受影响版本为 v3.0.0 到 v3.9.0,不包括 2.x 版本。
建议:
- 此 CVE 实际影响的项目与当前 Issue 所在项目不一致
- 建议关闭此 Issue,或将其移至正确项目的 Issue 列表
- 如有疑问,请联系安全团队确认
此评论由 Claude Agent Team(项目经理)自动生成


一、漏洞信息
漏洞编号:CVE-2025-9906
漏洞归属组件:python-Keras
漏洞归属的版本:2.10.0,2.12.0,2.3.1,2.4.3
CVSS评分:
BaseScore:7.3 High
Vector:CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:H/A:H
漏洞简述:
The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=True. One can create a specially crafted .keras model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special config.json (a file within the .keras archive) that will invoke keras.config.enable_unsafe_deserialization() to disable safe mode. Once safe mode is disable, one can use the Lambda layer feature of keras, which allows arbitrary Python code in the form of pickled code. Both can appear in the same archive. Simply the keras.config.enable_unsafe_deserialization() needs to appear first in the archive and the Lambda with arbitrary code needs to be second.
漏洞公开时间:2025-09-19 17:15:36
漏洞创建时间:2025-09-19 16:41:29
漏洞详情参考链接:
https://nvd.nist.gov/vuln/detail/CVE-2025-9906
更多参考(点击展开)
漏洞分析指导链接:
https://atomgit.com/openeuler/cve-manager/blob/master/cve-vulner-manager/doc/md/manual.md
漏洞数据来源:
七彩瞬析开源风险感知平台
漏洞补丁信息:
详情(点击展开)
二、漏洞分析结构反馈
影响性分析说明:
The Keras Model.load_model method can be exploited to achieve arbitrary code execution, even with safe_mode=True. One can create a specially crafted .keras model archive that, when loaded via Model.load_model, will trigger arbitrary code to be executed. This is achieved by crafting a special config.json (a file within the .keras archive) that will invoke keras.config.enable_unsafe_deserialization() to disable safe mode. Once safe mode is disable, one can use the Lambda layer feature of keras, which allows arbitrary Python code in the form of pickled code. Both can appear in the same archive. Simply the keras.config.enable_unsafe_deserialization() needs to appear first in the archive and the Lambda with arbitrary code needs to be second.
openEuler评分:
7.3
Vector:CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:U/C:H/I:H/A:H
受影响版本排查(受影响/不受影响):
1.master(2.12.0):受影响
2.openEuler-20.03-LTS-SP4:受影响
3.openEuler-22.03-LTS-SP3(2.10.0):受影响
4.openEuler-22.03-LTS-SP4(2.10.0):受影响
5.openEuler-24.03-LTS(2.12.0):受影响
6.openEuler-24.03-LTS-Next(2.12.0):受影响
7.openEuler-24.03-LTS-SP1(2.12.0):受影响
8.openEuler-24.03-LTS-SP2(2.12.0):受影响
9.openEuler-24.03-LTS-SP3(2.12.0):受影响
修复是否涉及abi变化(是/否):
1.master(2.12.0):否
2.openEuler-20.03-LTS-SP4:否
3.openEuler-22.03-LTS-SP3(2.10.0):否
4.openEuler-22.03-LTS-SP4(2.10.0):否
5.openEuler-24.03-LTS(2.12.0):否
6.openEuler-24.03-LTS-Next(2.12.0):否
7.openEuler-24.03-LTS-SP1(2.12.0):否
8.openEuler-24.03-LTS-SP2(2.12.0):否
9.openEuler-24.03-LTS-SP3(2.12.0):否
原因说明:
1.master(2.12.0):暂不修复-暂无解决方案或补丁
2.openEuler-20.03-LTS-SP4:暂不修复-暂无解决方案或补丁
3.openEuler-22.03-LTS-SP3(2.10.0):暂不修复-暂无解决方案或补丁
4.openEuler-22.03-LTS-SP4(2.10.0):暂不修复-暂无解决方案或补丁
5.openEuler-24.03-LTS(2.12.0):暂不修复-暂无解决方案或补丁
6.openEuler-24.03-LTS-Next(2.12.0):暂不修复-暂无解决方案或补丁
7.openEuler-24.03-LTS-SP1(2.12.0):暂不修复-暂无解决方案或补丁
8.openEuler-24.03-LTS-SP2(2.12.0):暂不修复-暂无解决方案或补丁
9.openEuler-24.03-LTS-SP3(2.12.0):暂不修复-暂无解决方案或补丁