Cybercriminals use ChatGPT as bait to spread malware, Kaspersky warns

Russian cybersecurity firm Kaspersky has reported a sharp rise in online scams where criminals disguise malware as popular artificial intelligence tools. Speaking at the St. Petersburg International Economic Forum, company representatives revealed that since the start of 2026, they have detected more than 92,000 cyberattacks worldwide that use fake AI services to trick users into downloading malicious software.

ChatGPT was the most common lure, accounting for 49 percent of all identified attacks. Claude and Gemini each made up 18 percent. In addition, Kaspersky analysts flagged nearly 15,000 samples of malicious software disguised as agent-based AI programs. These included banking trojans, spyware, exploits, and downloaders designed to install additional harmful modules.

Dmitry Galov, head of Kaspersky GReAT in Russia, warned that attackers are increasingly capitalizing on the popularity of AI services to steal personal data and money. “As cyber threats evolve rapidly, reliable security solutions have become essential to modern digital life,” he said.

To stay safe, Kaspersky advises using AI services from reputable companies, steering clear of anonymous or unfamiliar bots, and installing security software on all devices to block phishing sites and prevent malware installation.

The warning follows a separate alert from Russian authorities about the risks of using default passwords, which remains a leading cause of account and device compromises.