ASR AI Security Radar
Back to homepage

Recent AI Security Incidents

This archive includes all published incident pages. Page 6 of 18.

Each page is intended to help a security team answer three questions quickly: why the issue is AI-relevant, what part of the workflow may be exposed, and what actions should happen first.

Selection criteria and correction policy are documented in Methodology & Editorial Policy.

AI security incident: 2025-040: Critical Vulnerability in Windows Server Update Service (WSUS)

Incident date: Unknown | Published: March 4, 2026

On October 23, 2025, Microsoft released an out-of-band update to address a critical vulnerability in Windows Server Update Service (WSUS).

Read details

AI security incident: 2025-037: Multiple Vulnerabilities in F5 Products

Incident date: Unknown | Published: March 4, 2026

On October 15, 2025, F5 disclosed that a sophisticated nation-state actor breached its systems and maintained long-term persistent access into F5's infrastructure.

Read details

AI security incident: 2025-036: Critical Vulnerabilities in Cisco ASA and FTD

Incident date: Unknown | Published: March 4, 2026

On September 25, 2025, Cisco released several security advisories addressing 3 vulnerabilities, 2 of which are critical.

Read details

AI security incident: CVE-2026-27966 (NVD)

Incident date: February 26, 2026 | Published: February 26, 2026

Langflow is a tool for building and deploying AI-powered agents and workflows. Prior to version 1.8.0, the CSV Agent node in Langflow hardcodes allow dangerous code=True, which automatically exposes LangChain’s Python REPL tool (python...

Read details

AI security incident: Fickling has safety check bypass via REDUCE+BUILD opcode sequence (GHSA-mhc9-48gj-9gp3)

Incident date: February 25, 2026 | Published: February 25, 2026

Assessment It is believed that the analysis pass works as intended, REDUCE and BUILD are not at fault here. The few potentially unsafe modules have been added to the blocklist (https://github.

Read details

AI security incident: Fickling: OBJ opcode call invisibility bypasses all safety checks (GHSA-mxhj-88fx-4pcv)

Incident date: February 24, 2026 | Published: February 25, 2026

Assessment The interpreter so it behaves closer to CPython when dealing with OBJ , NEWOBJ , and NEWOBJ_EX opcodes (https://github.com/trailofbits/fickling/commit/ff423dade2bb1f72b2b48586c022fac40cbd9a4a).

Read details

AI security incident: Statamic is vulnerable to account takeover via password reset link injection (GHSA-jx...

Incident date: February 24, 2026 | Published: February 25, 2026

Impact An attacker may leverage a vulnerability in the password reset feature to capture a user's token and reset the password on their behalf.

Read details

AI security incident: Fiber has a Denial of Service Vulnerability via Route Parameter Overflow (GHSA-mrq8-r...

Incident date: February 24, 2026 | Published: February 25, 2026

A denial of service vulnerability exists in Fiber v2 and v3 that allows remote attackers to crash the application by sending requests to routes with more than 30 parameters.

Read details

AI security incident: Fiber is Vulnerable to Denial of Service via Flash Cookie Unbounded Allocation (GHSA-...

Incident date: February 24, 2026 | Published: February 25, 2026

Summary The use of the fiber_flash cookie can force an unbounded allocation on any server. A crafted 10-character cookie value triggers an attempt to allocate up to 85GB of memory via unvalidated msgpack deserialization.

Read details

AI security incident: Fiber has an Arbitrary File Read in Static Middleware on Windows (GHSA-m3c2-496v-cw3v)

Incident date: February 24, 2026 | Published: February 25, 2026

Summary **Description** A Path Traversal (CWE-22) vulnerability in Fiber allows a remote attacker to bypass the static middleware sanitizer and read arbitrary files on the server file system on Windows.

Read details