Fake Anthropic AI Preview Phishing Targets Enterprise Logins

What happened

Cybersecurity firm Cofense has documented a phishing campaign that targets enterprise employees with fake invitations to a private, invite-only preview of Anthropic's latest AI technology. The messages reportedly borrow GoTo Webinar branding, and the attackers aim to capture Google and Microsoft login credentials.

The lure relies on the appeal of early access. Recipients who are keen to test new AI tools are led to believe they have been selected for a restricted preview, when the invitation is in fact spoofed. Computerworld, which reported on the Cofense findings, notes that the attackers chose a brand that was prominent in the news at the time.

Frank Dickson of Dickson Research, quoted in the same coverage, described the tactic as a play on exclusivity and fear of missing out. He pointed out that organisations are still working out where AI fits into their operations, which makes a preview invitation especially tempting for ambitious staff. He also warned that the same approach could be applied to whichever AI name is dominating headlines next.

Why it matters

The campaign shows how AI adoption is creating new attack surfaces. As organisations rush to evaluate and deploy AI, employees face pressure to find and try new tools quickly, and that urgency is what a convincing invitation exploits. Google and Microsoft accounts sit at the centre of many enterprise workflows, so a single compromised login can expose far more than one person's mailbox.

For leaders driving digital transformation, the lesson is that AI rollout programmes shape behaviour as much as they shape technology. Where early-access routes are unclear, staff will fill the gap with whatever arrives in their inbox.

The Renascence take

This is less a story about Anthropic than about scarcity as a lever. Attackers do not need sophisticated malware when they can make a sign-up feel like a privilege, and exclusivity lowers the scrutiny that people apply to a message.

Most readers will file this under cybersecurity and move on, but the more useful question is why a plausible invitation beat a sceptical inbox. Scarcity and fear of missing out short-circuit verification, so awareness training that only warns against suspicious links will underperform. Operators should give staff one clearly announced channel for AI pilots, so that any invitation arriving outside it looks wrong at a glance. Make the legitimate path the easiest path, and the lure loses most of its power.