FounderIntellidark
Jeremy Makowski on criminal AI-as-a-Service in 2026
In research published by Rapid7 on 11 June 2026, Intellidark's founder maps how the underground market packages AI for fraud, impersonation and routine attack work.
Rapid7 published on 11 June 2026 "Criminal AI-as-a-Service in 2026: How the Underground Market Is Operationalizing Cybercrime", authored by Jeremy Makowski, founder of Intellidark.
The research finds that criminals mainly use AI as a productivity layer, speeding up routine work such as phishing, target profiling and translation, rather than as autonomous hacking. The underground market copies commercial packaging, with subscriptions and delivery over Telegram, but it is unstable and many offerings are repackaged public models. Prices range from free or cheap entry-level access to a few hundred dollars per month for FraudGPT-style services and thousands of dollars for professional-grade systems.
Two markets stand out as underestimated: stolen AI accounts and hijacked API keys, which multiply an attacker's capability and can expose sensitive prompts and files, and deepfake services used for impersonation and to bypass know-your-customer (KYC) checks. The article's advice to defenders: manage AI accounts and keys as critical assets, and rely on behavioural signals, identity verification and transaction controls, since polished language is no longer a reliable sign of fraud.
The full article is on Rapid7's blog; it is also listed on Intellidark's Research and Publications page.
Sources
Links open the source site. Figures are quoted as the sources give them.