The Role of Artificial Intelligence in Carding: Threat Evolution in 2026

[Good Carder]

Artificial intelligence (AI), particularly generative models (GenAI), has radically changed the landscape of carding — credit card fraud. By 2026, AI has become a key tool for cybercriminals, enabling them to automate processes, generate compelling content, create synthetic identities, and evade fraud detection systems. According to reports from Trend Micro, Group-IB, and Visa, 2025–2026 will mark the "fifth wave of cybercrime," with weaponized AI increasing the effectiveness of attacks.

The Russian-speaking underground, traditionally dominant in carding, has actively adopted AI: forums have developed sections on AI and machine learning, and mentions of "AI Agent" increased by 477% in 2025. This has led to an explosive growth in credit card fraud: experts estimate that GenAI is the cause of the "crime wave" in the credit card industry.

Generative AI for creating fake data and content​

One of GenAI's primary roles is generating convincing fake materials for phishing and social engineering.

• Personalized phishing emails and websites: Criminals use models like WormGPT or similar ChatGPT to create error-free text tailored to the victim. This makes spear phishing more convincing.
• Synthetic Identities: GenAI creates complete profiles (fullz) — name, address, SSN, credit history. This facilitates synthetic identity fraud, where fake identities are used to open accounts and cards.
• Card Number Generation: While banks use GenAI to predict compromised cards, criminals are experimenting with similar approaches to brute-force or generate valid BINs.

On Russian-language forums, carders share tutorials on using open-source models (including uncensored versions based on Grok or Mistral) for automation.

Process automation: AI agents and bots​

By 2026, carding will become "fully industrialized" thanks to agentic AI — autonomous agents that execute chains of actions.

• Automatic card checking: ML algorithms test thousands of dumps per second, identifying valid cards with minimal risk of blocking. AI agents imitate human behavior, bypassing CAPTCHA and rate limits.
• Autonomous transactions: Agents perform purchases, social engineering, and data extraction without human intervention. Visa notes a growing discussion of such tools in the underground.
• Dark AI tools: Specialized tools (WormGPT, Agent Zero, Business Invoice Swapper) adapted for fraud without ethical restrictions are sold on the darknet.

Trend Micro predicts that by 2026, such agents will make attacks fully autonomous.

Deepfakes and voice cloning for vishing​

Generative AI amplifies voice phishing (vishing) and video scams.

• Voice deepfake: Cloning the victim's voice or that of a bank manager to call support and lift restrictions.
• Deepfake videos: To bypass KYC (Know Your Customer) requirements when opening accounts. Group-IB has recorded a 40–52% increase in deepfake services on Telegram in 2025.

This is especially true when carding for cashing out via drops or mules.

Bypassing fraud detection systems​

Carders use ML to analyze and bypass banking anti-fraud systems.

• Pattern analysis: AI studies how detectors work (e.g. based on transactions) and generates "clean" patterns of behavior.
• Adversarial attacks: Models that fool banks' ML detectors (similar to how banks use GenAI to simulate attacks).

As a result, fraud has become "highly sophisticated": fewer attacks, but each one more effective.

Trends in the Russian-language underground​

Russian-language forums (XSS, Verified, and similar) are actively integrating AI:

• AI, Machine Learning & Automation sections with tutorials on bots and agents.
• Using open and "dark" models for scripting, cashing out and log analysis.
• Adapting to sanctions: AI aids anonymization and crypto transactions.

Group-IB and Trend Micro note that the Russian-language underground is evolving faster thanks to access to resources.

Forecasts and risks for 2026​

Experts predict further growth:

• GenAI will accelerate impersonation attacks and cyber fraud.
• Agentic AI will make carding autonomous and scalable.
• Banks are responding with their own AI systems, but the arms race continues.

Conclusion​

AI has transformed carding from a routine craft into a high-tech industry. Generative models and agents enable attacks to be carried out faster, cheaper, and more efficiently, increasing losses for banks and victims. This trend will only intensify in 2026, requiring financial institutions to invest in advanced AI detectors. The Russian-speaking underground remains at the forefront, adapting new technologies for maximum profit.