Artificial intelligence is fundamentally reshaping the cybersecurity landscape, introducing both unprecedented defensive capabilities and alarming new attack vectors.

The rapid evolution of generative AI and large language models (LLMs) has created a technological inflection point where digital identity verification mechanisms-including voice recognition, facial analysis, and behavioral biometrics-have become increasingly unreliable against sophisticated AI-driven impersonation attacks.

Account verification and unlocking service advertisement (Source – Check Point)

The cybersecurity ecosystem now confronts a paradigm where synthetic identities can be generated on demand, eroding traditional digital trust frameworks that organizations have relied upon for decades.

These technologies have dramatically increased the scalability and plausibility of social engineering campaigns, which remain among the most effective techniques for breaching security perimeters.

Check Point researchers have identified a troubling trend where cybercriminals are leveraging AI to streamline and enhance multiple stages of the malware kill chain, from generating ransomware scripts and phishing kits to building info-stealers and deepfakes.

Their analysis reveals this AI-assisted approach significantly lowers the barrier to entry for malicious actors and enables rapid scaling of attacks-even by technically unskilled individuals.

The implications extend beyond individual security incidents to a fundamental reshaping of enterprise risk profiles.

Check Point’s GenAI Protect data indicates AI services are used in at least 51% of enterprise networks monthly, with approximately 1 in 80 prompts (1.25%) posing a high risk of sensitive data leakage.

AI-Enabled Malware Development

The emergence of AI-assisted malware creation represents one of the most concerning developments in the current threat landscape.

A recent dark web discovery by Check Point researchers documented an attacker using ChatGPT to refine code for extracting credentials directly from Windows event logs.

The malicious actor was observed refining Windows API calls through LLM interaction to improve their credential harvesting capabilities.

EVT_HANDLE EvtQuery(
    [in] EVT_HANDLE Session,
    [in] LPCWSTR Path,
    [in] LPCWSTR Query,
    [in] DWORD Flags
);

This code snippet illustrates how attackers leverage the Windows Event Log API to extract potentially sensitive credential information.

Beyond code generation, AI is being deployed for post-attack data processing.

Threat actors are using specialized tools like DarkGPT-a malicious LLM modeled after ChatGPT-to analyze infostealer logs via natural language queries, enabling rapid extraction of credentials, domain names, API keys, and session tokens.

The efficiency gains provided by AI tools dramatically accelerate the attack lifecycle. By automating post-exfiltration analysis, attackers can quickly identify high-value targets for credential stuffing, account takeover, financial fraud, or enterprise intrusion-shortening the path to devastating ransomware deployment.

This operational shift represents a significant evolution in how cybercriminals leverage technology to maximize both efficiency and impact.

Are you from the SOC and DFIR Teams? – Analyse Malware Incidents & get live Access with ANY.RUN -> Start Now for Free.

The post Unmasking AI in Cybersecurity – From Dark-Web Tactics to Next-Gen Defenses appeared first on Cyber Security News.