Users on underground hacking and fraud forums are publicly complaining that a surge of low‑quality, AI‑generated posts is degrading discussion, depressing traffic and complicating moderation and intelligence work.
Members of underground hacking and fraud forums have begun to complain openly about an influx of AI‑generated content they say is low quality and harmful to community interactions. Anonymous threads on sites such as Hack Forums carry lines like "Stop posting AI shit" and "I’m disappointed that you are working to incorporate AI garbage into the site," reflecting growing irritation among scammers, grifters and low‑level hackers who rely on these boards for information and reputation building.
Researchers behind a recent study quantified the trend by examining 97,895 AI‑related conversations on cybercrime forums from ChatGPT's 2022 launch until the end of last year. The research team-which includes Ben Collier and scholars from the University of Edinburgh, the University of Cambridge and the University of Strathclyde — found many threads composed of short, "bullet‑pointed explainers" about basic cybersecurity topics and frequent complaints about low‑quality posts swamping discussion boards.
Forum structure and social norms help amplify the backlash. These message boards, often with historical roots in Russian‑language communities, function as both business platforms and social spaces where stolen data is traded, hacking jobs are advertised and reputation matters. Forum owners even run writing contests, so users value human‑to‑human interaction and visible signs of expertise; AI‑generated explainers can distort who appears knowledgeable and trusted.
The content shift is having measurable effects on traffic and engagement. The study cites concerns that AI search overviews are surfacing answers externally and reducing visits to forums, while participants say canned AI posts reduce incentives to write original content and make it harder to evaluate trustworthiness. That combination complicates both internal moderation and outside intelligence‑gathering, hollowing out the signal analysts and community moderators depend on.
On the technical side, interest in AI for criminal use remains real and varied. Actors have used AI for realistic face‑swapping, automated social engineering messages, code generation and vulnerability discovery. Security practitioners quoted in the reporting — including Ian Gray of Flashpoint — warn of two tiers of abuse: low‑skill actors amplifying noise with generic AI outputs, and more advanced adversaries who can bypass model safeguards and produce more convincing, harmful material.
For builders and defenders, the researchers and industry sources identify practical challenges and responses. AI‑driven noise erodes signal for forum moderation and threat intelligence; search‑overview features can hollow out traffic patterns investigators rely on; and adversaries are already experimenting with jailbreaks and translation‑enabled social engineering. Monitoring content quality, adapting reputation systems and hardening model guardrails are among the measures highlighted as necessary to protect both community integrity and investigative visibility.
Sources
Replies (0)
No replies in this topic yet.
