Google has released an open-source tool called PhishGuard AI that uses machine learning to detect phishing emails generated by artificial intelligence. The tool analyzes linguistic patterns, metadata, and behavioral signals to identify AI-crafted social engineering attempts with what Google claims is 94 percent accuracy.
The release comes as AI-generated phishing has become the fastest-growing category of email-based attacks. According to Google's Threat Analysis Group, the volume of AI-crafted phishing emails increased 280 percent in 2025, with attackers using large language models to create highly convincing messages that bypass traditional spam filters.
PhishGuard AI is available on GitHub under the Apache 2.0 license and can be integrated with major email platforms including Gmail, Outlook, and open-source mail servers. Google says the tool will receive regular model updates as AI phishing techniques evolve, and the company encourages the security community to contribute to the project.
