Researchers at Check Point2 and ESET3 have described two cases in which artificial intelligence was used in ransomware development or execution, though ESET said one sample was likely a proof of concept rather than malware used in attacks.

The reports matter because they put technical detail behind a broad claim about AI and ransomware. They document two specific cases; they do not measure how common AI-assisted ransomware is.

Check Point said FunkSec emerged publicly in late 2024 and claimed more than 85 victims in December2, more than any other ransomware group that month by its count. ESET said PromptLock was the “first known AI-powered ransomware”3 it had identified, but also said it was not spotted in real attacks and was likely a proof of concept or work in progress. Researchers flagged both because they suggest AI can lower the skill needed to build or automate parts of ransomware.

Check Point's report, whose publication date is not stated in the retrieved document, focused on FunkSec’s public activity and malware samples. The company said analysis of the group’s tools indicated likely AI-assisted development2 by “relatively inexperienced” operators, which it said may have contributed to rapid iteration.

According to Check Point, FunkSec’s custom ransomware was written in Rust, used the “.funksec” extension and appeared to be under active development, with versions published only days apart. Check Point said the operators also uploaded parts of the source code, including a file named ransomware.rs, and that several samples and related uploads pointed to development activity likely tied to Algeria.

The report also said FunkSec used double-extortion tactics, combining data theft with encryption, and sometimes demanded relatively low ransoms, “sometimes as little as $10,000.” Check Point added that many of the group’s leaked datasets appeared recycled from earlier hacktivist campaigns, which it said raised doubts about the authenticity of some disclosures and about using public victim claims as a proxy for impact.

Check Point further tied the group’s public persona to hacktivist branding. It said an allegedly leaked Trump-Netanyahu call promoted by a FunkSec-linked actor was “clearly AI-generated.”

ESET’s case is narrower but more explicit about how the AI component worked. In a post updated Sept. 3, 20253, ESET said PromptLock used OpenAI’s gpt-oss-20b model locally through the Ollama API3 to generate malicious Lua scripts during execution, then ran those scripts to enumerate the local filesystem, inspect target files, exfiltrate selected data and perform encryption.

ESET said PromptLock was written in Golang and that it identified both Windows and Linux variants uploaded to VirusTotal, listing sample hashes in the post. The company said the malware had the ability to exfiltrate, encrypt and “possibly even destroy data,” but added that this last function appeared not to be implemented3 in the samples it analyzed.

ESET also said PromptLock “was not spotted in actual attacks”3 and was instead thought to be a proof of concept or work in progress. In the Sept. 3 update, ESET said the authors of an academic study, Ransomware 3.0: Self-Composing and LLM-Orchestrated3, had contacted the company and said their research prototype closely resembled the PromptLock samples, which ESET said supported its view that the malware was not an operational campaign.

A separate Proofpoint press release dated July 22, 20264 addressed a different part of the problem: how victims perceived AI’s role in attacks, not malware authorship. Proofpoint said 65% of organizations affected by ransomware in its survey believed AI made the attack more effective, based on responses from 953 cybersecurity professionals across 12 countries, while also quoting Ryan Kalember as saying that “AI hasn't fundamentally changed ransomware, but it has materially improved the attacks that lead to ransomware.”

The Proofpoint release includes no independent test results. Its findings concern phishing, impersonation and credential theft, rather than direct evidence that ransomware code itself was generated by AI.

Taken together, the two research posts support a narrower conclusion than a claim of widespread autonomous ransomware. Check Point described a group whose tooling was likely built with AI assistance, while ESET documented samples that used a local language model to generate malicious scripts but said they were likely a proof of concept.