// Alert

Claude threat report

// ClaudeMEDIUM

Island researchers, as reported by GBHackers, documented a campaign dubbed TypoPrompting that uses lookalike domains of AI brands to deliver malware and browser hijackers. Oleg Zaytsev traced 186 lookalike domains across 27 brands, 18 of them imitating Claude, including typos such as clsude[.]ai, cluade[.]ai and caude[.]ai. Visitors are routed through traffic distribution systems that serve a parked page to most users and, for selected victims, a fake Cloudflare verification page that pastes a PowerShell download command into the Windows Run dialog. Eight Claude, DeepSeek, Copilot, ElevenLabs and Voice AI typos led to a browser-hijacking package named Net Guard. The campaign is not tied to a vulnerability in Claude itself, and the reported activity is based on research published November 13, 2026.

// Get alerts for Claude