What You Need
Before diving into the step-by-step breakdown, ensure you have a foundational understanding of cybersecurity concepts (threat actors, zero-days, malware, supply chain attacks), basic familiarity with generative AI and large language models (LLMs), and an awareness of recent threat intelligence reports (e.g., Google Threat Intelligence Group publications). No specialized tools are required—this guide is designed for security professionals, researchers, and informed readers seeking to grasp how adversaries operationalize AI.
Step-by-Step Guide
Step 1: Threat Actor Uses AI for Zero-Day Discovery and Mass Exploitation
For the first time, Google Threat Intelligence Group (GTIG) identified a criminal threat actor who developed a zero-day exploit with AI assistance. The adversary planned a mass exploitation event, but proactive counter-discovery by GTIG may have prevented its use. Nation-state actors linked to the People’s Republic of China (PRC) and the Democratic People’s Republic of Korea (DPRK) have also shown strong interest in leveraging AI for vulnerability discovery. This step underscores how AI lowers the barrier to finding and weaponizing unknown flaws.
Step 2: Accelerate Development of Polymorphic Malware with AI-Augmented Coding
Adversaries now use AI-driven coding to speed up the creation of infrastructure suites and polymorphic malware. AI enables the generation of obfuscation networks and integration of decoy logic, helping malware evade detection. Suspected Russia-nexus threat actors have been linked to such AI-enhanced development cycles. This step highlights defense evasion as a primary goal—AI allows rapid iteration of malware variants that can change their signature on the fly.
Step 3: Deploy Autonomous Malware for Attack Orchestration
AI-enabled malware like PROMPTSPY signals a shift toward autonomous operations. These malicious programs interpret system states and dynamically generate commands to manipulate victim environments. GTIG’s analysis revealed previously unreported capabilities, including the offloading of operational tasks to AI for scaled, adaptive activity. This step shows how adversaries can reduce human intervention and increase attack speed and complexity.
Step 4: Exploit AI as a High-Speed Research Assistant for Information Operations
Adversaries treat AI as a research assistant to support the full attack lifecycle. More importantly, they are moving toward agentic workflows to operationalize autonomous attack frameworks. In information operations (IO), AI generates synthetic media and deepfake content at scale to fabricate consensus. The pro-Russia IO campaign “Operation Overload” exemplifies this—AI tools amplified false narratives and created digital deception.
Step 5: Obtain Obfuscated, Premium-Tier Access to LLMs
Threat actors pursue anonymized, premium-tier access to large language models through professionalized middleware and automated registration pipelines. These methods bypass usage limits, enabling large-scale abuse of services like ChatGPT or Gemini. Adversaries also subsidize operations via trial abuse and programmatic account cycling. This step details the infrastructure-building phase that supports all other AI-driven activities.
Step 6: Target AI Environments Through Supply Chain Attacks
Groups like “TeamPCP” (aka UNC6780) have begun targeting AI environments and software dependencies as an initial access vector. These supply chain attacks compromise libraries, plugins, or cloud services used in AI pipelines, leading to multiple types of further compromise. This step closes the loop: once adversaries gain footholds through AI-adjacent weaknesses, they can pivot to broader espionage or disruption.
Tips
- Stay informed: Regularly review threat intelligence reports from GTIG, Mandiant, and other trusted sources to keep ahead of evolving AI-enabled tactics.
- Harden supply chains: Vet third-party AI components and dependencies rigorously—supply chain attacks are a growing vector.
- Monitor for obfuscated access: Watch for unusual patterns of API usage from anonymizing services; implement rate limiting and behavioral analytics.
- Invest in AI-empowered defense: Use AI for threat detection and response to match adversary speed—defense must evolve as fast as offense.
- Educate teams: Ensure your security team understands AI-generated malware and deepfakes; conduct tabletop exercises simulating autonomous attacks.
- Collaborate: Share intelligence on AI threats across sectors—adversaries share tools, defenders should share countermeasures.