IBM just caught the first malware written entirely by AI
IBM X-Force discovered 'Slopoly,' a ransomware tool likely generated by a large language model. It signals a new era where hackers use AI to build attacks faster than ever.
IBM's cybersecurity team just discovered something security experts have been warning about for years: malware that was almost certainly written by AI. The program, dubbed "Slopoly," was found inside a real ransomware attack — not a lab experiment — marking one of the first confirmed cases of AI-generated malware used in the wild.
The discovery, published by IBM X-Force on March 12, 2026, reveals how a criminal group called Hive0163 used Slopoly to maintain secret access to a company's servers for over a week before deploying ransomware that locked down systems and stole data.
How They Know AI Wrote It
IBM researchers found several telltale signs that a large language model (the technology behind ChatGPT, Claude, and similar AI tools) generated the malicious code:
- Overly detailed comments — the code was annotated like a tutorial, with explanations of every function. Human hackers almost never do this.
- Perfectly named variables — every variable had a clear, descriptive name, a pattern typical of AI-generated code but rare in hand-written malware.
- Built-in error handling — professional-grade logging and error messages that suggest the code was generated in a single structured session.
- False advertising — the code's own comments claimed it could change its own structure to avoid detection ("polymorphic" behavior), but it actually couldn't. The AI made promises the code didn't deliver.
- Unused functions — the AI included a timing function that was never actually called, suggesting iterative AI generation rather than human editing.
IBM researcher Golo Mühr concluded: "The quality suggests it was produced by a less advanced model" — but that's exactly what makes it alarming. Even a basic AI can now produce functional malware.
The Five-Stage Attack That Used It
Slopoly wasn't acting alone. It was part of a sophisticated five-step attack chain:
1. The trick — Hackers used "ClickFix" (a social engineering tactic where a fake error message convinces you to paste and run a command on your computer)
2. First backdoor — A program called NodeSnake quietly connected the victim's machine to the hackers' control server
3. Second backdoor — A more powerful tool called InterlockRAT added encrypted communication tunnels
4. Slopoly — The AI-written tool checked in with hackers every 30 seconds, sending system info and waiting for commands
5. Ransomware — After a week of surveillance, Hive0163 stole data and locked everything down
Why "Sloppy" AI Malware Is Actually Scarier
Security experts might be tempted to dismiss Slopoly — it's not sophisticated. It can't modify itself. It uses basic techniques. But IBM's assessment is chilling: "AI-generated malware does not pose a new or sophisticated threat from a technical standpoint. It disproportionately enables threat actors by reducing the time an operator needs to develop and execute an attack."
In plain English: the danger isn't that AI makes better malware — it's that AI lets more people make malware, faster. A criminal who once needed weeks to write an attack tool can now generate one in minutes. And because each version is freshly generated, traditional antivirus software that relies on recognizing known code patterns may not catch it.
Slopoly joins a growing list of suspected AI-generated threats including VoidLink and PromptSpy, signaling what IBM calls "the initial phase of an emerging arms race between adversarial AI and defenders."
How to Protect Yourself
IBM X-Force recommends several steps that apply to both businesses and individuals:
- Never paste commands from error messages — the "ClickFix" trick that started this attack relies on convincing you to copy-paste code you don't understand. If a website tells you to press Win+R and paste something, close the tab immediately.
- Use behavior-based security tools — antivirus that only recognizes known threats won't catch AI-generated malware. Look for tools that monitor what programs actually do, not just what they look like.
- Watch for unusual scheduled tasks — Slopoly hid as a Windows task called "Runtime Broker" (a real Windows process name). If you see unfamiliar scheduled tasks, investigate.
The era of AI-assisted cybercrime isn't coming — it's here. And the first real example was, ironically, kind of sloppy.
Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News
Stay updated on AI news
Simple explanations of the latest AI developments