MITRE ATT&CK v20 Tracks AI Abuse, MFA Bypass, Water Attacks

MITRE last week released version 20 of its ATT&CK framework, adding entries on artificial intelligence abuse, multi-factor authentication bypasses and attacks on water utilities in the United States and Mexico.

The update expands coverage of adversary activity across enterprise, industrial control systems and mobile environments. MITRE described the main change as “expanded coverage of recent adversary activity and tradecraft.”

The Enterprise domain has nine new techniques and sub-techniques. Two entries cover prompt injection: one involves malicious instructions entered directly into a large language model interface, while the other involves instructions hidden in documents or other content processed by the model. Another technique covers attackers using an organization’s AI agents to steal data.

Additional entries cover poisoning web content to influence AI search results and assistant recommendations, and disrupting the use of large language models in malware analysis. A separate technique tracks the theft of application-specific passwords, which can let attackers sign in without triggering an MFA prompt.

ATT&CK v20 adds a technique for bypassing CAPTCHA checks, with sub-techniques for automatically solving CAPTCHAs and avoiding them. Coverage of compromised software dependencies and development tools now includes configuration files used by integrated development environments and coding agents. MITRE described the additions as documenting how attackers get around “Multi-Factor Authentication, bot checks, and AI systems.”

The Industrial Control Systems domain adds a campaign covering attacks on internet-exposed programmable logic controllers at more than 30 water and wastewater utilities in Minnesota. The intrusions left operators unable to access their systems. Attribution remains unconfirmed, although the broader activity has been associated with Iranian-affiliated actors.

A second ICS campaign documents an AI-assisted breach of Mexican government institutions that affected information technology and operational technology systems and reached a municipal water utility. The ICS section adds the Command and Scripting Interpreter technique, replacing the deprecated Command-Line Interface and Scripting techniques. Six interpreter-specific sub-techniques are included, along with entries covering command-and-control traffic hidden in DNS and publish/subscribe protocols.

MITRE added Nimbus Manticore, an espionage group associated with Iran’s Islamic Revolutionary Guard Corps that targets aerospace, defense and telecommunications organizations, along with the group’s tools. The entry for China-backed Salt Typhoon now covers the exploitation of network edge devices. The APT38 entry covers software supply-chain attacks in addition to financial theft. MITRE also added The Gentlemen, a ransomware-as-a-service group.

Defenders receive 15 new detection strategies, including six for Enterprise and nine for ICS, supported by 38 analytics. The new ICS analytics are designed to exclude routine engineering and maintenance activity from detection results.

In the Mobile domain, MITRE restored a technique for delivering malicious applications outside official app stores. The technique last appeared in version 11. The update adds NGate, malware that relays payment card data through a phone’s near-field communication capability, allowing attackers to withdraw money from automated teller machines.