Artificial intelligence systems are now enabling attackers to automate the entire hacking process in real time, fundamentally disrupting the cybersecurity industry built over the past two decades, according to Alex Stamos, chief security officer at Cognition and a longtime cybersecurity expert. Open-weight AI models have advanced to the point where they can automatically identify software vulnerabilities, write exploits, and break into targets simultaneously, collapsing what was traditionally a multi-step process requiring teams of skilled hackers.
Stamos described a wave of breaches in recent weeks affecting major technology companies, including a chain of attacks that began with a compromise of Fireworks, an AI services provider. Attackers stole API keys from Fireworks that enabled them to breach Cursor, a code editor, where they pushed a malicious plugin to 5,000 additional companies, creating what Stamos called a "permanent continuous breach" across the tech industry. He also cited the Binance hack, which he characterized as the largest heist in history, and attacks on roughly 200 companies using ARTEX, a Chinese open-source hacking tool that anyone can download.
The shift eliminates barriers that previously limited cyberattacks. Where breaking into systems once required rare technical expertise and large teams, now attackers can use locally run AI models on consumer hardware to conduct sophisticated operations. Stamos said he has been running an AI model on an Apple Mac Studio that can reverse engineer software patches and write exploit code, capabilities that previously belonged only to elite hackers. Both criminal groups and state actors in countries including Russia, China, and North Korea are deploying these AI-powered tools.
Stamos compared the current moment to the 1990s, when the internet was new and security practices were still being developed, calling it a return to "Wild West days." The multibillion-dollar cybersecurity industry that emerged in the 2000s and 2010s was built on assumptions about skill hierarchies and operational structures that AI has now rendered obsolete. While Stamos said he is not pessimistic long-term, he believes the industry faces a "dark age" as it struggles to adapt to automated, AI-enabled attacks. }