AI 网络犯罪工具在地下市场的扩散

发生了什么

Halcyon 勒索软件研究中心的一份报告记录了地下网络犯罪市场中销售非法 AI 工具、越狱提示词、被盗账户以及 WormGPT 等恶意模型的激增现象。在 Telegram 频道、暗网论坛和专业市场中，此类广告从 2025 年末的每月不足 50 条增长到 2026 年 2 月的超过 1,400 条，这降低了网络攻击、身份欺诈和自动化诈骗基础设施的成本。

Commercialization and commodification of illicit AI tools, compromised accounts, and jailbreak services lower the barrier to entry for cybercrime and fraud across sectors.

证据摘录

• Advertisements for AI tools on underground platforms surged from under 50 per month in late 2025 to over 1,400 by February.
• Halcyon studied nearly 4,000 underground posts across 77 Telegram channels, 20 dark web forums, and five markets.
• Underground sellers offered compromised ChatGPT Plus accounts, jailbreak prompts, phishing tools, and malicious models like WormGPT.
• Scam tools sold in these markets include AI infrastructure capable of calling up to 120 people simultaneously.

严重度维度

来源引用

1. AI tools flood into underground cybercrime marketsAxios · 2026-10-06