As the ransomware industry evolves, experts predict that hackers will continue to find more and more ways to use this technology to exploit businesses and individuals.
Seksan Mongkol Kamsao | Moment | Getty Images
palo alto networks Technology director Lee Kralich said hackers are increasingly using artificial intelligence models to exploit vulnerabilities, leaving businesses with less time to strengthen their software defenses.
“We currently estimate that organizations have a narrow window of three to five months to outperform their adversaries before AI-powered exploits begin to become the new norm,” he said in a blog post Wednesday. “This impending flood of vulnerability demands urgency.”
The rise of increasingly sophisticated AI models, such as Anthropic’s Mythos, is increasing risk and putting pressure on cybersecurity teams to harden their defenses in preparation for a wave of cyberattacks that can exploit previously unknown software vulnerabilities. Concerns prompted the White House to hold a meeting with banking executives and technology giants.
google Although the government announced this week that it had halted attempts to use AI for “massive exploitation events,” hackers are already using available AI tools to exploit software vulnerabilities.
Klarich agreed that these capabilities are not limited to new models and called for industry-wide innovation to track new attack techniques, including virtual patching capabilities. He said Palo Alto will roll out the first feature set “soon.”
Last month, Anthropic limited the rollout of its Mythos model to select companies to test and fix vulnerabilities before hackers find and exploit them. This group included Palo Alto Networks. cloud strike, Amazon, apple and JP Morgan.
OpenAI announced the GPT-5.5-Cyber model last week, followed by the rollout of its Daybreak cyber initiative.
“The big question just a few weeks ago was, ‘Are we exaggerating the capabilities of the model?’ After further testing, we can confidently say that this is not the case,” Kralich wrote. “In fact, these models may be even better at finding vulnerabilities than we originally realized.”
