Live CVE & 0day intelligence

Sound the depths of every vulnerability.

Newly released CVEs across every platform — sleek to read, verbose on demand, and served raw as markdown for AI and agent ingestion. Severity reads as depth: the deeper the contact, the graver the threat.

Browse the feed For agents → /llms.txt

Depth = severity + exploitation

HadalCritical and actively exploited (CISA KEV / 0day)
AbyssalCritical with a public exploit or in-the-wild use
MidnightCritical, or high with PoC / in-the-wild
TwilightHigh severity, or a signal on a lesser flaw
SunlitLow / medium · no exploitation signal

69890

Tracked CVEs

6521

Critical

745

Exploited in wild

3413

Last 7 days

Trending now

Bluesky + Hacker News mentions, last 7 days

1CVE-2026-109618Critical· 9.2RCE via OOM corruption4896
2CVE-2026-109619Critical· 9.2RCE via stale global state2196
3CVE-2026-111497Critical· 9.2Out-of-Bounds Write in Google Brotli Decoder1792
4CVE-2026-99999Critical· 9.3In the Linux kernel, the following vulnerability has been resolved: ipv4: validate checksum_start before completing checksum If a packet with bad checksum metadata gets into the ipv4 stack, skb_checksum_help can corrupt the network hea…963
5CVE-2026-110264Critical· 10.0Integer Overflow or Wraparound in GitLab367
6CVE-2026-91403Critical· 9.1Vulnerability in the Oracle Java SE, Oracle GraalVM for JDK, Oracle GraalVM Enterprise Edition product of Oracle Java SE (component: JSSE)262

This week's digest →All digests

Feed

17 matches

CVE-2026-12570Medium· 5.5

2mo ago

A vulnerability in keras-team/keras versions <= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function

A vulnerability in keras-team/keras versions <= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function. The H5IOStore.__getitem__ method in keras/src/saving…

▾ Sunlitkeras · kerasEPSS 0.19%via NVD

CVE-2026-12484High· 7.8

3mo ago

Keras versions prior to 3.14.0 are vulnerable to a path traversal issue in the archive extraction utilities located in `keras/src/utils/f…

Keras versions prior to 3.14.0 are vulnerable to a path traversal issue in the archive extraction utilities located in `keras/src/utils/file_utils.py`. The functions `filter_safe_tarinfos()` and `filter_safe_zipinfos()` validate archive …

▾ Twilightkeras · kerasEPSS 0.56%via OSV

CVE-2026-1462High· 7.8

6mo ago

A vulnerability in the `TFSMLayer` class of the `keras` package, version 3.13.0, allows attacker-controlled TensorFlow SavedModels to be loaded during deserialization of `.keras` models, even when `safe_mode=True`

A vulnerability in the `TFSMLayer` class of the `keras` package, version 3.13.0, allows attacker-controlled TensorFlow SavedModels to be loaded during deserialization of `.keras` models, even when `safe_mode=True`. This bypasses the secu…

▾ Twilightkeras · kerasEPSS 0.40%via NVD

CVE-2026-1669High· 7.5

8mo ago

Arbitrary file read in the model loading mechanism (HDF5 integration) in Keras versions 3.0.0 through 3.13.1 on all supported platforms allows a remote attacker to read local files and disclose sensitive information via a crafted .keras …

Arbitrary file read in the model loading mechanism (HDF5 integration) in Keras versions 3.0.0 through 3.13.1 on all supported platforms allows a remote attacker to read local files and disclose sensitive information via a crafted .keras …

▾ Twilightkeras · kerasEPSS 0.33%via NVD

CVE-2026-0897High· 7.5PoC

9mo ago

Allocation of Resources Without Limits or Throttling in the HDF5 weight loading component in Google Keras 3.0.0 through 3.13.0 on all platforms allows a remote attacker to cause a Denial of Service (DoS) through memory exhaustion and a c…

Allocation of Resources Without Limits or Throttling in the HDF5 weight loading component in Google Keras 3.0.0 through 3.13.0 on all platforms allows a remote attacker to cause a Denial of Service (DoS) through memory exhaustion and a c…

▾ Midnightkeras · kerasEPSS 0.33%via NVD

CVE-2025-12060Critical· 9.8

11mo ago

Keras Directory Traversal Vulnerability

▾ Midnightkeras · kerasEPSS 0.63%via OSV

CVE-2025-12058Medium

1y ago

Keras is vulnerable to arbitrary local file loading and Server-Side Request Forgery

▾ Sunlitkeras · kerasEPSS 0.25%via OSV

CVE-2025-49655Critical· 9.8

1y ago

Keras framework vulnerable to deserialization of untrusted data

▾ Midnightkeras · kerasEPSS 0.75%via OSV

CVE-2025-9905High

1y ago

The Keras `Model.load_model` method silently ignores `safe_mode=True` and allows arbitrary code execution when a `.h5`/`.hdf5` file i…

The Keras `Model.load_model` method **silently** ignores `safe_mode=True` and allows arbitrary code execution when a `.h5`/`.hdf5` file is loaded.

▾ Twilightkeras · kerasEPSS 0.22%via OSV

CVE-2025-9906High· 7.3

1y ago

Keras is vulnerable to Deserialization of Untrusted Data

▾ Twilightkeras · kerasEPSS 0.20%via OSV

CVE-2025-8747High· 8.8

1y ago

Keras vulnerable to CVE-2025-1550 bypass via reuse of internal functionality

▾ Twilightkeras · kerasEPSS 0.12%via OSV

CVE-2025-1550HighPoC

1y ago

Arbitrary Code Execution via Crafted Keras Config for Model Loading

▾ Midnightkeras · kerasEPSS 2.6%via OSV

CVE-2024-55459Medium

1y ago

keras Path Traversal vulnerability

▾ Sunlitkeras · kerasEPSS 0.24%via OSV

From the blog

All posts →

Oct 27, 2026

Sound the depths of every vulnerability.

Trending now

Feed

A vulnerability in keras-team/keras versions <= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function

Keras: TorchModuleWrapper can deserialize unsafe PyTorch pickle data

Keras: tar extraction permits symlink-based path traversal

Keras: HDF5 virtual datasets can disclose local files

Keras: DiskIOStore permits path traversal through crafted layer names

Keras versions prior to 3.14.0 are vulnerable to a path traversal issue in the archive extraction utilities located in `keras/src/utils/f…

A vulnerability in the `TFSMLayer` class of the `keras` package, version 3.13.0, allows attacker-controlled TensorFlow SavedModels to be loaded during deserialization of `.keras` models, even when `safe_mode=True`

Arbitrary file read in the model loading mechanism (HDF5 integration) in Keras versions 3.0.0 through 3.13.1 on all supported platforms allows a remote attacker to read local files and disclose sensitive information via a crafted .keras …

Allocation of Resources Without Limits or Throttling in the HDF5 weight loading component in Google Keras 3.0.0 through 3.13.0 on all platforms allows a remote attacker to cause a Denial of Service (DoS) through memory exhaustion and a c…

Keras Directory Traversal Vulnerability

Keras is vulnerable to arbitrary local file loading and Server-Side Request Forgery

Keras framework vulnerable to deserialization of untrusted data

The Keras `Model.load_model` method **silently** ignores `safe_mode=True` and allows arbitrary code execution when a `.h5`/`.hdf5` file i…

Keras is vulnerable to Deserialization of Untrusted Data

Keras vulnerable to CVE-2025-1550 bypass via reuse of internal functionality

Arbitrary Code Execution via Crafted Keras Config for Model Loading

keras Path Traversal vulnerability

From the blog

VulnSea has moved to vulnsea.com (and what changed since September)

What's new: open corpus, three new MCP tools, vendor advisories, package pages

Wiring VulnSea into an agent: a hands-on tutorial

Sound the depths of every vulnerability.

Trending now

Feed

A vulnerability in keras-team/keras versions <= 3.15.0 allows for a denial of service (DoS) attack when loading malicious .keras model files via the keras.models.load_model() function

Keras: TorchModuleWrapper can deserialize unsafe PyTorch pickle data

Keras: tar extraction permits symlink-based path traversal

Keras: HDF5 virtual datasets can disclose local files

Keras: DiskIOStore permits path traversal through crafted layer names

Keras versions prior to 3.14.0 are vulnerable to a path traversal issue in the archive extraction utilities located in `keras/src/utils/f…

A vulnerability in the `TFSMLayer` class of the `keras` package, version 3.13.0, allows attacker-controlled TensorFlow SavedModels to be loaded during deserialization of `.keras` models, even when `safe_mode=True`

Arbitrary file read in the model loading mechanism (HDF5 integration) in Keras versions 3.0.0 through 3.13.1 on all supported platforms allows a remote attacker to read local files and disclose sensitive information via a crafted .keras …

Allocation of Resources Without Limits or Throttling in the HDF5 weight loading component in Google Keras 3.0.0 through 3.13.0 on all platforms allows a remote attacker to cause a Denial of Service (DoS) through memory exhaustion and a c…

Keras Directory Traversal Vulnerability

Keras is vulnerable to arbitrary local file loading and Server-Side Request Forgery

Keras framework vulnerable to deserialization of untrusted data

The Keras `Model.load_model` method **silently** ignores `safe_mode=True` and allows arbitrary code execution when a `.h5`/`.hdf5` file i…

Keras is vulnerable to Deserialization of Untrusted Data

Keras vulnerable to CVE-2025-1550 bypass via reuse of internal functionality

Arbitrary Code Execution via Crafted Keras Config for Model Loading

keras Path Traversal vulnerability

From the blog

VulnSea has moved to vulnsea.com (and what changed since September)

What's new: open corpus, three new MCP tools, vendor advisories, package pages

Wiring VulnSea into an agent: a hands-on tutorial

The Keras `Model.load_model` method silently ignores `safe_mode=True` and allows arbitrary code execution when a `.h5`/`.hdf5` file i…

The Keras `Model.load_model` method silently ignores `safe_mode=True` and allows arbitrary code execution when a `.h5`/`.hdf5` file i…