Anthropic, OpenAI's Next Models Could Be A 'Watershed' Event For Cybersecurity, Warns Expert—'Agentic Attackers Are Coming'

Industry experts are warning of rising cybersecurity risks from AI agents, saying these tools can identify and exploit system vulnerabilities faster and more persistently than human hackers.

Shlomo Kramer, founder and CEO of cybersecurity and networking company Cato Networks, told CNN, “The agentic attackers are coming. This is a watershed event in the history of cybersecurity.”

Experts warn that AI can both intensify existing risks and accelerate the creation of new software exploits, with Kramer cautioning that each new generation of models will bring more serious cybersecurity threats.  

“Behind Mythos is the next OpenAI model, and the next Google Gemini from Alphabet (NASDAQ:GOOG) (NASDAQ:GOOG), and a few months behind them are the open-source Chinese models,” he said.

AI Threats Trigger Cybersecurity Concerns

This warning comes on the heels of reports of a data leak at Anthropic on Wednesday, which inadvertently exposed internal data about an unreleased model, “Claude Mythos”. The company described this model as a “step change” in capabilities, with significantly better performance in “reasoning, coding, and cybersecurity.” Shares of CrowdStrike Holdings Inc (NASDAQ:CRWD), Zscaler Inc (NASDAQ:ZS) and iShares Expanded Tech-Software Sector ETF (BATS:IGV) declined following the report.

OpenAI had also raised concerns in December about its upcoming models posing a “high” cybersecurity risk.

Meanwhile, Nikesh Arora, CEO of Palo Alto Networks (NASDAQ:PANW), warned that these AI models could boost cyberattacks within six months and could be easily accessible to anyone with a credit card and a computer. He noted that AI-powered attacks could go from breach to data theft in just 25 minutes, while most firms take days to detect intrusions.

George Ralph, Global Managing Director and Chief Revenue Officer (CRO) at RFA, echoed the same and told Benzinga that AI is heightening cybersecurity risks for private equity firms, making attacks more sophisticated and enabling even less-skilled hackers to carry out advanced scams.

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Image via Shutterstock