Hackers Dislike AI Garbage Even More Than You Do

The complaint feels recognizable. “I’m frustrated that you are planning to integrate AI nonsense into the site,” one irritated user, posting anonymously, expressed in an online message. “No one is asking for this—we want you to enhance the site, not charge us for new features.”

However, this isn’t a typical internet user lamenting about AI being thrust into their favorite app. Instead, the concerns come from a cybercrime forum regarding the introduction of more generative AI. Like millions of others, scammers, con artists, and low-tier hackers are expressing dissatisfaction over AI’s intrusion into their activities and the influx of low-quality AI content flooding their online spaces.

“People are unhappy,” states Ben Collier, a security researcher and senior lecturer at the University of Edinburgh. In a recent study examining how low-level cybercriminals utilize AI, Collier and his colleagues noted a growing resistance to the adoption of generative AI within underground cybercrime forums and hacking collectives.

Throughout the waves of hype surrounding generative AI in the last few years, some participants on hacking forums have shifted from positivity about AI’s potential to greater skepticism regarding its value, according to the study, which also included researchers from the University of Cambridge and the University of Strathclyde.

The researchers scrutinized 97,895 AI-related discussions on cybercrime forums from the launch of ChatGPT in 2022 to the end of the previous year. They uncovered grievances about users posting “bullet-pointed explainers” of basic cybersecurity principles, complaints concerning the surplus of low-quality contributions, and worries about Google’s AI search summaries reducing traffic to the forums.

For decades, cybercrime message boards and marketplaces, often originating from Russia, have facilitated collaboration among scammers. They serve as platforms for trading stolen data, advertising hacking jobs, and where fraudsters indulge in banter about their competitors. While scammers sometimes attempt to con one another, the forums foster a sense of community. For instance, users cultivate reputations for reliability, and forum moderators conduct writing contests.

“These are basically social hubs. They really detest the use of [AI] on the forums,” Collier observes. He notes that the social dynamics can be disrupted when aspiring cybercriminals try to enhance their reputations by sharing AI-generated hacking tutorials. “I believe many are somewhat conflicted about AI since it undermines their claim to expertise.”

Content reviewed by WIRED on Hack Forums, a venue for those interested in discussing hacking and sharing techniques, indicates frustration over AI-generated posts. “I notice many members using AI to create their threads/posts, and it annoys me as they can’t even take the time to write a few lines,” one user commented. Another stated more succinctly: “Stop posting AI garbage.”

In several cases, Collier mentions, participants across multiple forums seem to be irritated by AI content as they seek genuine social interactions. “If I wanted to engage with an AI chatbot, there are plenty of sites for that … I come here for human connection,” one post referenced in the research reads.

Since the arrival of ChatGPT in late 2022, there has been considerable interest in AI’s hacking potential and how the technology might reshape online crime. Both skilled hackers and those with fewer abilities have been exploring ways to leverage AI in their attacks. While some organized criminals have enhanced their schemes using increasingly realistic AI face-swapping technologies and AI-translated social engineering messages, much of the focus has been on generative AI’s abilities to craft malicious code and identify vulnerabilities.