AI Threat Intelligence

Threat intelligence, automated.

AI-researched threat write-ups, live IOC feeds, and MITRE ATT&CK mappings — human-reviewed before every publish. Built for defenders.

Threats tracked—
IOCs published—
MITRE techniques—
Last updated—
Latest Report
HIGHphishing

Device Code Phishing Risks for AI Coding Agents

Lab research indicates that device code phishing, a technique targeting device-flow clients, can be extended to compromise AI coding agents like GitHub Copilot CLI. This method allows attackers to gain unauthorized access to developer accounts by tricking users into authenticating malicious applications, posing a significant risk to development environments.

Read full report →
Recent Threats
CRITICALvulnerability

CVE-2026-90436: Apache Axis 1.x Unauthenticated Remote Code Execution

Apache Axis 1.x contains a critical unsafe reflection vulnerability (CVE-2026-90436) allowing unauthenticated remote code execution. Attackers can craft SOAP fault messages to instantiate arbitrary classes available on the application classpath. This vulnerability affects all Axis 1.x versions through 1.4, which are end-of-life, and is actively exploited in the wild.

Read report →
CRITICALvulnerability

CVE-2024-10441: Synology DSM/BeeStation OS Remote Code Execution

CVE-2024-10441 is a critical improper encoding vulnerability in Synology DiskStation Manager (DSM) and BeeStation OS (BSM) that allows remote attackers to execute arbitrary code. With a CVSS score of 9.8, this vulnerability poses a significant risk to unpatched Synology devices and is listed in CISA’s KEV catalog. Immediate patching is strongly recommended.

Read report →
CRITICALvulnerability

CVE-2026-90817: REDCap Unauthenticated Remote Code Execution

A critical unauthenticated remote code execution (RCE) vulnerability, CVE-2026-90817, has been identified in Vanderbilt University’s Research Electronic Data Capture (REDCap) software versions 13.3.0 and higher. This code injection flaw allows attackers to execute arbitrary code on the server by manipulating HTTP requests, provided they know a valid public survey hash. CISA has added this CVE to its Known Exploited Vulnerabilities Catalog.

Read report →
HIGHvulnerability

CVE-2026-40217: BerriAI LiteLLM Remote Code Execution via Unprotected Channel

CVE-2026-40217 is a high-severity remote code execution vulnerability in BerriAI LiteLLM through version 2026-04-08. Exploitation occurs via bytecode rewriting at the `/guardrails/test_custom_code` URI, allowing remote attackers to execute arbitrary code. This vulnerability is actively exploited, often associated with “LLMjacking” attacks to monetize compromised cloud and model resources.

Read report →
HIGHthreat

OPTIMJACK Supply Chain Attack Leverages Compromised Vendor Script

The OPTIMJACK supply chain attack involves a compromised third-party vendor script that, when embedded in legitimate websites, can lead to their exposure. This highlights the critical need for robust client-side security measures and strict segmentation to prevent unauthorized data access or manipulation.

Read report →
CRITICALvulnerability

Critical Puppet Server RCE (CVE-2026-115731) Under Active Exploitation

CISA has added CVE-2026-115731, a critical remote code execution (RCE) vulnerability in Perforce Puppet Server, to its Known Exploited Vulnerabilities catalog. This flaw, with a CVSS score of 9.9, allows an authenticated client to execute arbitrary code on the server, with unauthenticated attackers potentially gaining access via weak certificate autosigning. Active exploitation has been reported.

Read report →
CRITICALvulnerability

CVE-2026-115731: Perforce Puppet Enterprise Deserialization RCE

A critical deserialization of untrusted data vulnerability (CVE-2026-115731) in Perforce Puppet Enterprise’s report-processing code allows arbitrary code execution. An attacker with a CA-signed agent certificate can exploit this by submitting a crafted report via a PUT request. This vulnerability poses a significant risk, especially where certificate authorities automatically sign requests without strong verification, enabling unauthenticated network attackers to gain a valid certificate.

Read report →
HIGHapt

QUIC and Covert: Novel HTTP/3-Powered RATs Target Southeast Asian Governments

A cyberespionage campaign is targeting Southeast Asian governments using custom Remote Access Trojans (RATs), `PerfectPioneer` and `KUKURAT`. This campaign is notable for its use of HTTP/3 (QUIC) for command and control (C2) communications, enhancing stealth and evasion capabilities against traditional network defenses.

Read report →

Transparently AI-authored

Every report on this site is researched and drafted by an AI agent, then reviewed and approved by a human analyst before publication. The Agent Logbook shows every step — sources consulted, enrichment calls, tokens used, and approval status — in real time.