AI-driven cyberattacks will start to be the 'new norm' in months, Palo Alto warns

According to Lee Klarich, the chief technology officer at Palo Alto Networks, companies are rapidly running out of time to enhance their software defenses against the growing threat of AI-driven cyberattacks. In a recent blog post, he emphasized that organizations now have a limited window of just three to five months to improve their security measures before these attacks become commonplace. Klarich warned that hackers are increasingly leveraging advanced AI models, such as Anthropic's Mythos, to exploit vulnerabilities in software. This alarming trend has prompted urgent discussions at the White House involving banking leaders and tech industry executives. He highlighted that as cybercriminals gain access to sophisticated AI tools, the potential for widespread exploitation of previously unknown vulnerabilities is on the horizon. In a recent incident, Google managed to thwart an attempt to conduct a mass exploitation event using AI, but the threat remains as hackers continue to take advantage of existing AI capabilities. Klarich stressed that the challenge is not limited to new AI models, urging the industry to innovate and develop new techniques for detecting and mitigating these cyber threats, including the implementation of virtual patching. Palo Alto Networks is poised to introduce a new set of capabilities aimed at enhancing cybersecurity soon. Last month, Anthropic limited its Mythos rollout to a select group of companies, including Palo Alto Networks, CrowdStrike, Amazon, Apple, and JPMorgan, to identify and address vulnerabilities before they could be exploited by cybercriminals. Meanwhile, OpenAI announced its GPT-5.5-Cyber model and the launch of its Daybreak cyber initiative, further highlighting the critical need for robust cybersecurity measures. Klarich concluded by reflecting on the capabilities of these models, asserting that they are more adept at identifying vulnerabilities than previously thought.