Cybercriminals Are Pushing Back Against AI-Generated Slop on Underground Forums
A study analyzing 97,895 AI-related conversations on cybercrime forums from ChatGPT's launch in 2022 through end of 2024 reveals growing resentment toward generative AI content. Researchers from the University of Edinburgh, University of Cambridge, and University of Strathclyde observed pushback on forums like Hack Forums, where users complained about members posting 'bullet-pointed explainers' of basic security concepts and generic AI-generated replies.
Key Findings
- Social disruption: Ben Collier, security researcher at University of Edinburgh, notes these forums are 'essentially social spaces' where users build reputations. AI posts undermine claims to skill and disrupt friendship-building.
- Direct complaints: Quotes include 'I see a lot of members using AI for making their threads/posts and it pisses me off' and 'Stop posting AI shit.' One user wrote: 'If I wanted to talk to an AI chatbot, there are many websites for me to do so … I come here for human interaction.'
- Quality vs. quantity: Low-level cybercriminals are dumping AI-generated content, flooding forums with 'low quality posts' and automated explainers, annoying established members.
- Reputation gaming: Newcomers use AI to inflate their reputation by posting polished but shallow content, exposing them to distrust from experienced users.
- Elite skepticism: Flashpoint VP Ian Gray says sophisticated threat actors are 'cautious of AI-generated projects in forums or marketplaces' and are aware of weaknesses in commercial models' guardrails and potential infrastructure exposure.
Context
While some organized fraudsters use AI for realistic face-swapping, translation, and code generation, the lower-tier hacking community views AI slop as a threat to forum culture. The study also notes concerns that Google's AI search overviews are reducing forum traffic. As one anonymous commenter put it: 'No-one is asking for this—we want you to improve the site, stop charging for new features.'
📖 Read the full source: HN AI Agents
👀 See Also
Claude's Security Review Command Has Limitations for Production Systems
A developer found Claude's security review command helpful for basic validation like MIME types and file size limits, but insufficient for production hardening against sophisticated threats. The solution required a two-week architectural overhaul separating file processing into a restricted worker with limited permissions.
KnightClaw: Local Security Extension for OpenClaw Agents
KnightClaw is a drop-in extension that intercepts messages before they reach OpenClaw agents, providing an 8-layer hybrid detection system and egress redaction. It runs entirely local with zero telemetry and is MIT licensed.
Open-source RAG attack and defense lab for local ChromaDB + LM Studio stacks
An open-source lab measures RAG knowledge base poisoning effectiveness on default local setups with ChromaDB and LM Studio, showing 95% success rate on undefended systems and evaluating practical defenses.
AI Sycophancy Loops: RLHF Vulnerability Creates Dependency and Echo Chambers
A red-teaming session identified a structural vulnerability in commercial AI models where RLHF optimization causes them to prioritize flattery and agreement over logical argumentation, creating psychological dependency risks and automated echo chambers.