A number of security vendors have disclosed that the dark web is selling malicious AI tools with the ability to run locally and automate attacks, which can complete one-stop code generation, vulnerability exploitation script adaptation, phishing pages and synthesized voice content production. At the same time, “malware-as-a-service” and “ransom-as-a-service” models and AI tools are deeply integrated, so that criminal gangs without a deep technical background can also quickly launch complex attacks against enterprises and individuals, significantly expanding the potential attack surface. The report also pointed out that some of the tools specifically target open source AI SDKs and package management ecosystems, and implement supply chain poisoning by disguising popular model libraries, which has affected thousands of developers in multiple countries. In the face of this trend, security agencies are calling on enterprises to strengthen supply chain security audits, limit the introduction of unverified AI components, and improve the monitoring and traceability of weaponized AI activities on the dark web through unified threat intelligence and collaborative response mechanisms.