PROMPTPurify

promptpurify LLMチャットアプリ向けの超軽量プロンプトインジェクションファイアウォール。約14MB。CPUのみ。 ユーザー入力とLLMの間にそのまま差し込めるガード——同じマシン上で動作し、 GPU不要、API不要、追加サービス不要。 SecureLayer7 レッドチームが開発。多くの OSSガードレールは数百MB級でGPUを要求し、それでも本番で見られる攻撃を 見逃します。私たちは自社のAI製品や顧客のアプリにそのまま組み込める、 そういった依存の一切ないものを必要としていました。 これが存在する理由 | promptpurify| 一般的なOSSガードレール...

EUVD-2026-77721

Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference toolkit. Prior to 0.21.17, 0.22.3, and 0.23.2, the tract-onnx crate passes the attacker-controlled externaldata location from an ONNX model through onnx/src/tensor.rs getexternalresources and joins the value to the model...

CVE-2026-55832

Tract (tract-onnx crate, a Rust inference toolkit for TensorFlow/ONNX) is vulnerable to arbitrary local file read when loading an untrusted ONNX model. The external_data location field is joined to the model directory via Path::join without rejecting absolute paths or .. components, so the defaul...

CVE-2026-55832 Tract: Arbitrary file read via unsanitized ONNX external_data `location` (path traversal) on model load in tract-onnx

Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference toolkit. Prior to 0.21.17, 0.22.3, and 0.23.2, the tract-onnx crate passes the attacker-controlled externaldata location from an ONNX model through onnx/src/tensor.rs getexternalresources and joins the value to the model...

CVE-2026-55832 Tract: Arbitrary file read via unsanitized ONNX external_data `location` (path traversal) on model load in tract-onnx

Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference toolkit. Prior to 0.21.17, 0.22.3, and 0.23.2, the tract-onnx crate passes the attacker-controlled externaldata location from an ONNX model through onnx/src/tensor.rs getexternalresources and joins the value to the model...

Cache Poisoning

Spring AI is vulnerable to cache poisoning. The vulnerability is due to insecure handling of the deterministic cache path, where an attacker can pre‑create the cache directory and plant a malicious ONNX model file and achieve execution when the model is loaded...

EUVD-2026-66822

A local attacker on a multi-user host can pre-create the deterministic cache path and plant a malicious ONNX model file. Spring AI 2.0.0 Spring AI 1.1.0 - 1.1.8 Spring AI 1.0.0 - 1.0.9...

CVE-2026-47852

A local attacker on a multi-user host can pre-create the deterministic cache path and plant a malicious ONNX model file. Spring AI 2.0.0 Spring AI 1.1.0 - 1.1.8 Spring AI 1.0.0 - 1.0.9...

CVE-2026-47852: Insecure Temporary File

A local attacker on a multi-user host can pre-create the deterministic cache path and plant a malicious ONNX model file. Spring AI 2.0.0 Spring AI 1.1.0 - 1.1.8 Spring AI 1.0.0 - 1.0.9...

CVE-2026-47852 Predictable cache directory location allows local ONNX model substitution in Spring AI

A local attacker on a multi-user host can pre-create the deterministic cache path and plant a malicious ONNX model file. Spring AI 2.0.0 Spring AI 1.1.0 - 1.1.8 Spring AI 1.0.0 - 1.0.9...

CVE-2026-47852 Predictable cache directory location allows local ONNX model substitution in Spring AI

A local attacker on a multi-user host can pre-create the deterministic cache path and plant a malicious ONNX model file. Spring AI 2.0.0 Spring AI 1.1.0 - 1.1.8 Spring AI 1.0.0 - 1.0.9...

CVE-2026-47852

CVE-2026-47852 affects Spring AI (versions 1.0.0–1.0.9 , 1.1.0–1.1.8 , and 2.0.0 ). A local attacker on a multi-user host can exploit a predictable (deterministic) cache directory path by pre-creating it and planting a malicious ONNX model file , leading to integrity impact as rated by the vendor...

CVE-2026-47852 Predictable cache directory location allows local ONNX model substitution in Spring AI

A local attacker on a multi-user host can pre-create the deterministic cache path and plant a malicious ONNX model file. Spring AI 2.0.0 Spring AI 1.1.0 - 1.1.8 Spring AI 1.0.0 - 1.0.9...

PT-2026-82350

Name of the Vulnerable Software and Affected Versions Spring AI versions 1.0.0 through 2.0.0 Description A local attacker on a multi-user host can perform cache poisoning by pre-creating a deterministic cache path and planting a malicious ONNX Open Neural Network Exchange model file. ONNX is an...

cve-2026-47852 - HIGH - Predictable cache directory location allows local ONNX model substitution in Spring AI

cve-2026-47852 - HIGH - Predictable cache directory location allows local ONNX model substitution in Spring AI...

Insecure Temporary File

Overview Affected versions of this package are vulnerable to Insecure Temporary File via the predictable cache directory location. An attacker can substitute a malicious ONNX model file by pre-creating the deterministic cache path on a multi-user host. Remediation Upgrade...

PT-2026-59335

Summary Null pointer dereference SIGSEGV in Upsample 6 7::adapt upsample 6 7 onnx/version converter/adapters/upsample 6 7.h:31 when convert version processes a model with an Upsample node that has zero inputs. The adapter accesses node-inputs0-sizes without checking input count. 107-byte PoC...

CVE-2026-40979

In Spring AI, having access to a shared environment can expose the ONNX model used by the application. Affected versions: Spring AI: 1.0.0 - 1.0.5 fixed in 1.0.6, 1.1.0 - 1.1.4 fixed in 1.1.5...

Security Bulletin: IBM Watson Speech Services Cartridge is vulnerable to an Improper Input Validation in ONNX [CVE-2026-34445, CVE-2026-34446, CVE-2026-34447]

Summary IBM Watson Speech Services Cartridge is vulnerable to an Improper Input Validation in ONNX due to an issue with the ExternalDataInfo class in ONNX using Python's setattr function to load metadata like file paths or data lengths directly from an ONNX model file, which fails to properly...

Spring AI's ONNX model cache defaults to world-writable predictable /tmp directory

In Spring AI, having access to a shared environment can expose the ONNX model used by the application. Affected versions: Spring AI: 1.0.0 - 1.0.5 fixed in 1.0.6, 1.1.0 - 1.1.4 fixed in 1.1.5...