Security readout for executives and security teams
Plain-English summary
A malicious Keras model archive can make affected systems allocate excessive memory while loading weights, crashing the Python process. Business risk is service interruption in ML pipelines, model upload portals, notebooks, or automation that loads externally supplied .keras files.
Executive priority
Treat as high priority for ML services that process external models. It is primarily an availability risk, but a crash in shared ML infrastructure can interrupt customer workflows, batch processing, or internal analytics.
Technical view
CVE-2026-0897 is CWE-770 in Google Keras 3.0.0 through 3.13.0. The HDF5 weight-loading path may trust dataset shape metadata in a .keras archive, causing memory exhaustion and interpreter crash. CVSS is 7.6 with user interaction required.
Likely exposure
Exposure is most likely where Keras loads model archives from users, partners, CI jobs, model registries, or downloaded artifacts. Systems loading only internally built, trusted models have lower practical exposure.
Exploitation context
The bundle reports no CISA KEV listing and provides no evidence of active exploitation. Exploitation requires a crafted model archive to be loaded by a vulnerable Keras process, so social, workflow, or upload-path delivery matters.
Researcher notes
The source bundle is strong on vulnerability class and impact, but incomplete on exact fixed upstream version. The affected metadata also appears narrower than the narrative description, so verify against vendor advisories before final scoping.
Mitigation direction
- Inventory Keras versions and prioritize 3.0.0 through 3.13.0.
- Apply applicable Red Hat errata for packaged deployments.
- Check Keras vendor guidance for the fixed upstream release.
- Do not load untrusted .keras or HDF5 model artifacts.
- Run model-loading jobs with memory limits and process isolation.
Validation and detection
- Review dependency locks, containers, and notebooks for vulnerable Keras versions.
- Map services that accept or import external model archives.
- Confirm upload pipelines quarantine untrusted model artifacts before loading.
- Verify model-loading workers have enforced memory limits.
- Check monitoring for Python worker crashes during model import.
Public sources used
Generated from the cited source records. This long-tail analysis has not been individually reviewed by a named human.
Conservative CVE-to-ATT&CK context
These mappings and lookup hints may be relevant to the vulnerability behavior, CWE, affected product, or exposure path. Glexia-inferred context is not an official MITRE, ATT&CK, CWE, or CVE Program mapping.
ATT&CK lookup starting points
Use these exact CWE pages and searches to review the Glexia ATT&CK library from this CVE's weakness and description context.
CWE-770: Exact CWE lookup
Use the exact CWE identifier as the starting point before reviewing related ATT&CK behavior. Open the exact CWE lookup page first, then review the ATT&CK searches from that MITRE weakness context. This is a Glexia lookup hint, not an official ATT&CK mapping.
Open ATT&CK lookupCVE-2026-0897 mapping review
Open the CVE-to-ATT&CK bridge for reviewed, inferred, or future official mappings tied to this CVE.
Open ATT&CK lookup- Severity
- High
- CVSS
- 7.6 (3.1)
- Known Exploited
- No
- Published
Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:L/A:H
CNA and ADP enrichment extracted from CVE v5
These fields come from the CVE record and ADP containers, not from Glexia's Take. They preserve time-varying source decisions such as CISA SSVC, KEV status, CVSS metrics, and provider references.
SSVC decision data
CVSS vector scores
2 official scoresWe collect every scored CVSS vector available in the official CNA and ADP containers. When more than one version is present, the table keeps the source vectors side by side instead of collapsing them into the highest score.
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:L/A:H2.84.7redhat-SADPCVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N——GoogleVulnerability scoring details
Base CVSS 4.0 score
7.1HighVector: CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
Base CVSS 3.1 score
7.6HighVector: CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:L/A:H
Vulnerability timeline
Timeline events are normalized from CVE metadata, CNA source timelines, ADP timelines, and KEV metadata when present.
- CVE reservedCVE Program
The CVE ID was reserved by the assigning CNA.
- ADP timelineredhat-SADP
Made public.
- CVE publishedCVE Program
The CVE record was published.
- ADP timelineredhat-SADP
Reported to Red Hat.
- CVE updatedCVE Program
The CVE record metadata indicates this as the latest update time.
ADP provider summaries
- 2026-01-15T16:01:16.399Z: Reported to Red Hat.
- 2026-01-15T14:09:53.603Z: Made public.
Source materials
- CVE List V5 sourceCVE List V5
- https://github.com/keras-team/keras/pull/21880CVE reference, Google
- https://access.redhat.com/security/cve/CVE-2026-0897CVE reference, redhat-SADP · vdb-entry, x_refsource_REDHAT
- RHBZ#2430027CVE reference, redhat-SADP · issue-tracking, x_refsource_REDHAT
- https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-0897.jsonCVE reference, redhat-SADP · x_sadp-csaf-vex
- https://access.redhat.com/errata/RHSA-2026:3782CVE reference, redhat-SADP · vendor-advisory, x_refsource_REDHAT
- https://access.redhat.com/errata/RHSA-2026:3713CVE reference, redhat-SADP · vendor-advisory, x_refsource_REDHAT
- https://access.redhat.com/errata/RHSA-2026:4271CVE reference, redhat-SADP · vendor-advisory, x_refsource_REDHAT
Products and packages named in the record
CWE details
CWE links open Glexia weakness intelligence pages with official CWE context, developer remediation guidance, and related CVE mappings.
Allocation of Resources Without Limits or Throttling
Allocation of Resources Without Limits or Throttling represents a recurring weakness pattern that can create exploitable paths when design, validation, or implementation controls are missing.