AI-Powered Zero-Day Exploit Uncovered: Google Threat Intelligence Reports First-Ever Use by Criminal Adversaries

Breaking: Criminal Threat Actor Deploys AI-Developed Zero-Day Exploit

For the first time, Google Threat Intelligence Group (GTIG) has identified a criminal threat actor using a zero-day exploit that researchers believe was developed with artificial intelligence. The exploit was intended for a mass exploitation event, but GTIG's proactive counter-discovery may have prevented its use.

AI-Powered Zero-Day Exploit Uncovered: Google Threat Intelligence Reports First-Ever Use by Criminal Adversaries — Source: www.mandiant.com

This marks a significant escalation in adversarial AI adoption, moving from theoretical risk to real-world operational capability. According to GTIG, the actor had planned to leverage the exploit for widespread attacks before researchers intervened.

AI-Augmented Code Accelerates Defense Evasion

AI-driven coding is enabling adversaries to rapidly develop infrastructure suites and polymorphic malware. Russia-nexus threat actors have been linked to AI-generated decoy logic embedded in malware, making detection more difficult.

These AI-enabled development cycles allow for obfuscation networks that evade traditional security measures. Attackers can now iterate malware variants at machine speed, outpacing signature-based defenses.

Autonomous Malware: PROMPTSPY Ushers in New Era

GTIG has uncovered previously unreported capabilities in PROMPTSPY, an AI-enabled malware that autonomously orchestrates attacks. It interprets system states to dynamically generate commands and manipulate victim environments.

This malware signals a shift toward autonomous attack orchestration, where threat actors offload operational tasks to AI for scaled, adaptive activity. The implications for incident response are profound: attacks can now adapt in real-time without human intervention.

Background: The Maturing AI Threat Landscape

Since GTIG's February 2026 report on AI-related threats, adversaries have transitioned from nascent AI experiments to industrial-scale use of generative models. The current environment is dual-natured: AI serves both as a sophisticated engine for attacks and as a high-value target.

Threat actors associated with the People's Republic of China (PRC) and the Democratic People's Republic of Korea (DPRK) have shown significant interest in using AI for vulnerability discovery. Meanwhile, state-sponsored groups are integrating AI into every phase of the attack lifecycle.

Supply Chain Attacks Target AI Environments

Adversaries like TeamPCP (aka UNC6780) have begun targeting AI software dependencies as an initial access vector. These supply chain attacks can lead to multiple types of compromise, from data theft to model poisoning.

AI environments themselves are now prime targets. Attackers seek to infiltrate the very systems that power modern machine learning workflows, creating a cascading risk for organizations that rely on AI.

What This Means

The use of AI-developed zero-day exploits represents a paradigm shift in cyber threat intelligence. Traditional detection methods may struggle against exploits that were generated through AI's ability to find and weaponize vulnerabilities at machine speed.

Organizations must now prepare for an environment where adversaries operate with AI-augmented speed and scale. This includes investing in AI-driven defense systems and proactive counter-exploit measures.

GTIG urges immediate reassessment of supply chain risks, especially for AI infrastructure. The era of autonomous malware and AI-as-a-service for criminals has arrived, demanding urgent collaboration across public and private sectors.

Expert Quote

"This is a watershed moment. For the first time, we have concrete evidence of a threat actor using a zero-day exploit that was likely generated by AI, not manually coded," said a senior GTIG analyst. "Our proactive discovery may have prevented a mass exploitation event, but others will follow."

The analyst added that AI-augmented operations are now industrial-scale, and defenders must adapt accordingly.

Anonymized LLM Access Fuels Misuse

Threat actors are also pursuing anonymized, premium-tier access to large language models through professionalized middleware and automated registration pipelines. This infrastructure bypasses usage limits and enables large-scale misuse.

These operations are subsidized through trial abuse and programmatic account cycling, allowing criminals to maintain persistent access to AI tools. The trend underscores the need for stricter access controls and abuse detection by AI providers.

Key Developments at a Glance

Zero-Day Exploit: First confirmed AI-developed zero-day used by criminal actor; prevented by GTIG.
Polymorphic Malware: Russia-nexus actors leverage AI for obfuscation and decoy logic.
Autonomous Malware: PROMPTSPY demonstrates AI-driven attack orchestration.
Supply Chain Risks: AI environments targeted as initial access vectors.
LLM Abuse: Anonymized access and trial abuse sustain large-scale misuse.

For more details, see GTIG's full Background and What This Means sections.

Tags: