MITRE ATT&CK v20 formally adds AI abuse to its threat-tracking framework
MITRE released version 20 of its ATT&CK framework, the most widely used tool for cataloguing attacker tactics worldwide, adding nine new Enterprise techniques that explicitly cover AI abuse. Prompt injection is now split into two separate techniques, one for malicious instructions an attacker types directly into an LLM and another for instructions hidden in content the model simply reads, while other additions cover attackers abusing an organization's own AI agents to steal data and poisoning web content to skew AI search and assistant recommendations. The release adds 15 new detection strategies overall, six for Enterprise and nine for industrial control systems (ICS), backed by 38 analytics, and documents an AI-assisted breach of Mexican government institutions that reached a municipal water utility. MITRE says the biggest change in this release is its 'expanded coverage of recent adversary activity and tradecraft,' including how attackers bypass MFA and CAPTCHA checks. For a business building its own threat-detection program, folding AI abuse into such a widely used reference framework means related controls will start showing up soon in off-the-shelf security tools it already uses.
Sources
This summary was written automatically with AI from the public sources listed above and checked by an independent AI reviewer before publication. For full details, read the original source.