ASR AI Security Radar
Back to homepage

Recent AI Security Incidents

This archive includes all published incident pages. Page 3 of 16.

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: Fickling has always check safety() bypass: pickle.loads and pickle.loads remain unhoo...

Incident date: March 4, 2026 | Published: March 4, 2026

Fickling has always check safety() bypass: pickle.loads and pickle.loads remain unhooked Assessment The missing pickle entrypoints pickle.loads, pickle.loads, and pickle.load were added to the hook https://github.

Read details

AI security incident: jackson-core has Nesting Depth Constraint Bypass in UTF8DataInputJsonParser potential...

Incident date: March 4, 2026 | Published: March 4, 2026

jackson-core has Nesting Depth Constraint Bypass in UTF8DataInputJsonParser potentially allowing Resource Exhaustion Summary The UTF8DataInputJsonParser, which is used when parsing from a java.io.

Read details

AI security incident: Authlib: Setting alg: none and a blank signature appears to bypass signature verifica...

Incident date: March 4, 2026 | Published: March 4, 2026

Authlib: Setting alg: none and a blank signature appears to bypass signature verification Summary After upgrading the library from 1.5.2 to 1.6.0 (and the latest 1.6.

Read details

AI security incident: Craft CMS has unauthenticated activation email trigger with potential user enumeratio...

Incident date: March 4, 2026 | Published: March 4, 2026

Craft CMS has unauthenticated activation email trigger with potential user enumeration The actionSendActivationEmail() endpoint is accessible to unauthenticated users and does not require a permission check for pending users.

Read details

AI security incident: time calibrator was removed from crates.io due to malicious code (GHSA-77xj-rrh3-wx3v)

Incident date: March 4, 2026 | Published: March 4, 2026

time calibrator was removed from crates.io due to malicious code It was reported time calibrator contained malicious code, that would try to upload .env files to a server.

Read details

AI security incident: IRRd: web UI host header injection allows password reset poisoning via attacker-contr...

Incident date: March 4, 2026 | Published: March 4, 2026

IRRd: web UI host header injection allows password reset poisoning via attacker-controlled email links Impact An attacker can manipulate the HTTP Host header on a password reset or account creation request.

Read details

AI security incident: locutus call user func array vulnerable to Remote Code Execution (RCE) due to Code In...

Incident date: March 4, 2026 | Published: March 4, 2026

locutus call user func array vulnerable to Remote Code Execution (RCE) due to Code Injection Details A Remote Code Execution (RCE) flaw was discovered in the locutus project (v2.0.

Read details

AI security incident: Vaultwarden has Privilege Escalation via Bulk Permission Update to Unauthorized Colle...

Incident date: March 4, 2026 | Published: March 4, 2026

Vaultwarden has Privilege Escalation via Bulk Permission Update to Unauthorized Collections by Manager Summary A Manager account (access all=false) was able to escalate privileges by directly invoking the bulk-access API against...

Read details

AI security incident: OpenClaw Vulnerable to Local File Exfiltration via MCP Tool Result MEDIA: Directive I...

Incident date: March 4, 2026 | Published: March 4, 2026

OpenClaw Vulnerable to Local File Exfiltration via MCP Tool Result MEDIA: Directive Injection Summary A malicious or compromised MCP (Model Context Protocol) tool server can exfiltrate arbitrary local files from the host system by...

Read details

AI security incident: OpenClaw: Hardlink alias checks could bypass workspace-only file boundaries in specif...

Incident date: March 4, 2026 | Published: March 4, 2026

OpenClaw: Hardlink alias checks could bypass workspace-only file boundaries in specific configurations Summary In certain workspace-restricted configurations, OpenClaw could follow hardlink aliases inside the workspace that reference...

Read details