AI Model Theft and Roundcube Webmail Exploits Surge in March

Checkpoint’s March threat report warns of a coordinated campaign that combines two distinct attack vectors. Adversaries are creating fraudulent accounts on cloud platforms to download and exfiltrate large AI models, while a separate group is actively exploiting two publicly disclosed vulnerabilities in Roundcube Webmail to gain unauthorized access to email servers.

The theft of AI models threatens intellectual property, undermines competitive advantage, and can be used to accelerate further attacks. Exploiting Roundcube allows threat actors to harvest credentials, read sensitive communications, and pivot within the network. Defenders must prioritize detection of anomalous account provisioning, enforce strict access controls on AI assets, and immediately patch or mitigate the Roundcube flaws to block the webmail attack chain.

Categories: Vulnerabilities & Exploits, AI Security & Threats, Threat Intelligence

Source: Read original article