Phishing Campaign Exploits Anthropic AI to Steal Credentials

• •Phishing emails impersonating Anthropic target Google and Microsoft accounts.
• •Attackers use URL shorteners and CAPTCHAs to bypass email security tools.
• •The campaign has been active since September 2026, with ongoing spoofing tactics.

Since September 2026, attackers have been sending phishing emails impersonating Anthropic, aiming to steal Google and Microsoft account credentials. The emails, which promise exclusive access to a new AI tool, contain multiple red flags, including grammatical errors and incorrect sender information. Despite these obvious signs, the emails have bypassed leading secure email gateways due to the use of URL shorteners and CAPTCHAs. Victims who click the links are directed to a fake Zoom meeting page, where they may inadvertently provide their login credentials. Cofense has noted that the campaign's inept tactics raise questions about how it evaded spam filters. The threat actor has also begun spoofing Anthropic's website, indicating possible use of AI in the campaign. Organizations are urged to remain vigilant against such phishing attempts.