Loading story…
Security reporting has documented LameHug as the first publicly known malware to integrate a large language model, marking an escalation in the use of generative AI within malicious tooling. The development indicates that AI can be incorporated into malware operations rather than being used solely to create phishing content or assist attackers outside the payload.
Separately, ransomware victim listings reached 894 organizations in July 2026, according to NCC Group data cited by ZDNET, with industrial organizations comprising nearly one-third of listed victims. The reporting also identified the first documented fully agentic AI ransomware attack chain, attributed to JadePuffer, amid a surge led by groups including The Gentlemen and Qilin; however, organizations should treat leak-site claims cautiously because some actors, including the new CRPxO RaaS operation, may inflate or fabricate victim listings.
Track how attackers are adapting to this technology.
6 events from the most recent confirmed update back to the earliest known activity.
The first documented fully agentic AI ransomware attack chain was recorded in July 2026. The incident was attributed in the report to JadePuffer.
ExfilSquad claimed to have stolen 570,000 records from Analog Devices following an alleged July incident. The claim was not verified.
Coca-Cola subsidiary Fairlife suffered a ransomware attack in July that was thought to be the work of Anubis. The group claimed it stole more than 1 TB of data.
EY experienced a July data breach that exposed client information and tax records. ShinyHunters claimed responsibility for the incident.
The ransomware-as-a-service operation CRPxO emerged in July 2026 and quickly claimed 36 organizational compromises. NCC Group assessed the claims as low-to-moderate credibility because the group had not released datasets or obtained victim confirmation and presented inconsistent evidence.
NCC Group recorded 894 ransomware victim-organization listings in July 2026, a year-to-date monthly high and a 22% increase from June. The industrial sector represented nearly one-third of recorded incidents, while U.S. organizations accounted for 41% of listings.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
2 references tracked. Mallory keeps watching after this page renders.
Map indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Researchers reported that the **JadePuffer** campaign carried out a largely autonomous ransomware intrusion using an AI agent and a large language model to move from initial access to encryption with limited human involvement. The operation reportedly exploited **`CVE-2025-3248`** in Langflow for unauthenticated remote code execution, then conducted reconnaissance, credential theft, persistence, privilege escalation, and lateral movement before pivoting to a production server running **Alibaba Nacos**, where ransomware was deployed and a Bitcoin ransom note was displayed. Reporting also linked the activity to exploitation of **`CVE-2021-29441`** in Nacos during the attack chain. The campaign drew attention because the AI system reportedly adapted to failures on its own, correcting a failed login in about 31 seconds and rewriting payloads without operator intervention, demonstrating machine-speed intrusion behavior that could compress defender response time. At the same time, researchers said the extortion component appeared broken: the encryption key was effectively random and not retained, making recovery impossible even for victims willing to pay, and the Bitcoin wallet in the ransom note was reportedly fabricated. The case has been cited as evidence that AI-assisted ransomware has moved from theory to documented practice, even if the criminal monetization model in this instance was flawed.
Aug 6, 2026
Security researchers have identified a new trend in which threat actors are embedding large language models (LLMs) directly into malware to enhance its capabilities and evade detection. Akamai Hunt discovered a novel malware strain that disguises its command and control (C2) traffic as legitimate LLM API requests, using Base64-encoded strings to communicate and potentially allowing attackers full control over compromised systems and data exfiltration. This approach enables malicious traffic to blend in with normal AI-related network activity, making detection more challenging for defenders. Further analysis and industry reporting highlight that malware families such as PromptFlux and PromptSteal are now querying LLMs mid-execution to dynamically alter their behavior, obfuscate code, and generate system commands on demand. PromptFlux, for example, uses the Gemini API to regularly re-obfuscate its source code, while PromptSteal leverages the Hugging Face API for real-time reconnaissance and exfiltration commands. These developments underscore the need for organizations to adapt their security controls and detection strategies to address the evolving threat landscape where AI and LLMs are weaponized by attackers.
Jul 12, 2026
Malware developers are actively experimenting with artificial intelligence, specifically large language models (LLMs), to create adaptive malware capable of rewriting its own code during execution. Google Threat Intelligence Group has identified malware families such as PromptFlux and PromptSteal that utilize LLMs to dynamically generate, modify, and execute scripts, allowing these threats to evade traditional detection methods. PromptFlux uses Gemini's API to regularly mutate its VBScript payloads, issuing prompts like "Act as an expert VBScript obfuscator" to the model, resulting in self-modifying malware that continually alters its digital fingerprints. PromptSteal, meanwhile, masquerades as an image generator but leverages a hosted LLM to generate and execute one-line Windows commands for data theft and exfiltration, effectively functioning as a live command engine. These AI-driven malware samples are still considered experimental, with limited reliability and persistence compared to traditional threats, but they represent a significant evolution in attack techniques. Notably, PromptSteal was reportedly used by Russia-linked APT28 (also known as BlueDelta, Fancy Bear, and FROZENLAKE) against Ukrainian targets, marking the first observed use of LLMs in live malware operations. The emergence of purpose-built AI tools for cybercrime is lowering the barrier for less sophisticated actors, and researchers warn that the integration of AI into malware development could soon lead to more autonomous, adaptive, and harder-to-detect threats. Google has taken steps to disrupt these operations, but the trend signals a shift toward more unpredictable and rapidly evolving attack patterns.
Jun 29, 2026