The rise of AI-powered ransomware marks a new and dangerous chapter in cybersecurity, according to a recent warning issued by ESET researchers. The cybersecurity firm has identified what it describes as the first known ransomware strain that actively learns from each infected system, adapting its behavior in real time using artificial intelligence. The discovery signals a major shift in how cyberattacks are designed, executed, and defended against.
Unlike traditional ransomware, which relies on prewritten code and predictable execution paths, this new threat operates dynamically. Named PromptLock, the malware does not follow a fixed script.
Instead, it communicates with an AI model during an attack, generating and refining malicious code as it spreads. This ability to evolve on the fly makes it significantly harder to detect, analyze, and contain.
At the core of PromptLock is a dual-layer architecture. Its main controller, written in the Go programming language, acts as a command unit that connects to a remote AI system. Rather than carrying a full set of attack instructions, the malware contains embedded prompts that instruct the AI to generate Lua scripts when needed. These scripts are then executed on the victim’s machine, allowing the ransomware to adapt to different environments in real time.
This approach represents a fundamental shift in malware design. Instead of executing predefined behavior, PromptLock effectively “asks” an AI model how to proceed based on what it finds inside a system. As a result, the attack process becomes highly personalized, with each infected machine receiving unique malicious instructions tailored to its configuration and data.
Once inside a system, PromptLock begins by analyzing files, processes, and system structure. Based on its findings, it decides whether to encrypt data, exfiltrate sensitive information, or destroy files entirely. These decisions are not rule-based but generated dynamically by the AI model, making detection far more difficult for traditional security tools.
What makes the threat even more concerning is its built-in feedback mechanism. After executing a script, the malware collects execution logs and sends them back to the AI model.
If the script fails or produces errors, the model adjusts its output and generates a corrected version. This iterative loop allows the ransomware to improve its effectiveness during an active attack, mimicking a form of real-time learning.
Security researchers note that this capability significantly reduces one of the traditional weaknesses of malware, coding errors. In earlier ransomware families, a single bug could expose the attack or cause it to fail. With AI in the loop, PromptLock can self-correct, making it far more resilient and adaptive.
ESET’s report also highlights that PromptLock is not operating in isolation. Researchers identified related tools such as PromptFlux, which rewrites malware code to improve persistence, and PromptSteal, which generates Windows commands designed to locate and steal sensitive documents. Together, these tools form part of a growing ecosystem of AI-assisted cybercrime utilities.
This development comes amid a sharp rise in ransomware activity worldwide. According to ESET, publicly disclosed ransomware victims in 2025 have already exceeded last year’s totals, with a projected increase of nearly 40 percent by year-end. Established ransomware groups such as Qilin and Akira remain active, while newer players are adopting AI-driven techniques to evade detection and scale operations more efficiently.
The emergence of AI-powered ransomware represents a turning point for cybersecurity defenses. Traditional signature-based detection methods struggle against malware that changes its behavior for every victim. Even behavior-based tools face challenges when malicious actions closely resemble legitimate system operations generated dynamically by an AI model.
ESET researchers warn that this shift could fundamentally alter how ransomware campaigns are conducted. Instead of mass-distributed attacks using identical payloads, future threats may rely on smaller, highly adaptive deployments that are harder to trace, analyze, or stop. This evolution increases the burden on security teams and underscores the need for advanced behavioral monitoring, zero-trust architectures, and AI-powered defense systems.
The report also raises broader concerns about the dual-use nature of artificial intelligence. While AI continues to deliver major benefits in healthcare, productivity, and research, its misuse by cybercriminals demonstrates how easily the same technology can be weaponized. The ability to generate code, analyze environments, and adapt behavior in real time makes AI an especially powerful tool in the wrong hands.
ESET emphasizes that organizations must treat this development as a wake-up call. Strong endpoint protection, continuous monitoring, regular patching, and employee awareness remain critical. More importantly, defenders must begin adopting AI-driven security tools capable of matching the speed and adaptability of emerging threats.
As AI continues to reshape both innovation and cybercrime, the discovery of PromptLock signals a new era in digital security, one where malware no longer follows scripts but learns, adapts, and evolves on its own.
Stay ahead of emerging cyber threats and AI breakthroughs, visit ainewstoday.org for the latest updates shaping the future of technology.
