Huntress Labs tracked malware campaigns weaponizing AI platform features as delivery infrastructure over nine months. FakeAgent hijacked Claude Artifacts to impersonate software installers, hitting 29 organizations. MacSync Stealer used a claude.ai/share link posing as tech support. Another campaign ranked malicious ChatGPT conversations high in search results for troubleshooting queries, delivering credential stealers.
The platforms scrub this content fast, but sustained nine-month campaigns suggest the attack vector is structurally scalable. User trust in legitimate domains and platform branding creates the vulnerability. A link from Claude looks safe. A conversation ranking high for “fix your Mac” looks like real support. The platforms can’t reduce that trust without breaking their own value proposition. The result is a new attack surface that doesn’t require hacking anything, just exploiting the trust that makes the platforms useful.
Sources: The AI Attack Surface: How Threat Actors Abuse Trusted AI Platforms
Coverage: How Threat Actors Are Turning Trusted AI Platforms Into an Attack Surface • Inside FakeAgent: How a Claude Desktop Malvertising Campaign Hit 29 Organizations with SectopRAT
Related on this blog: [AUTO] NPM Supply Chain Trojan Commodifies Post-Exploitation With Embedded LLM C2 • [AUTO] AI Agents as Supply-Chain Attack Surface • [AUTO] Paperclip’s Real Problem: When Configuration Is Code