The cybersecurity industry is reeling from a wave of attacks that one prominent expert says marks the beginning of an AI-driven shift in how intrusions are carried out. Alex Stamos, chief security officer at Cognition and a lecturer at Stanford University's Cyber Policy Center, described a month of continuous breaches across the technology sector, with companies left exhausted by the pace of the response.
Speaking about the past few weeks, Stamos pointed to a series of incidents that were chained together. A breach at Fireworks, a provider of AI services, exposed API keys that attackers used to get into Cursor; from there they pushed a bad plugin to 5,000 more companies, creating 5,000 potential breaches. Other incidents included a Binance hack that Stamos described as the largest heist of anything in human history, and breaches at a series of AI-related neocloud companies.
Stamos attributed the escalation to AI enabling attackers to compress the cyber kill chain, the sequence of steps used to steal information, shut down systems or take money. Earlier, research, vulnerability discovery and exploit writing happened in advance before deployment against a target, but AI systems now find bugs and write exploits in real time, he said. A Chinese open-source tool called ARTEX, powered on the back end by something like an obliterated GLM-5.3, has been used to break into a striking number of organizations, starting with three South Korean banks and expanding to about 200 more companies this week, according to Stamos.
Stamos compared the current period to the 1990s, when he attended early DEFCON gatherings of a couple hundred people and attendees would leave talks convinced that every company in the world was vulnerable. The difference, he said, is that the internet did not underpin everything then, and attacks could not be automated as they are now. He noted that many people now have access to such tools, often live where law enforcement cannot reach them, and face strong economic incentives. Stamos also said he has run a quantized obliterated GLM-5.3 on an M5 Mac Ultra Studio and found its capabilities remarkable.
Stamos said both private and state actors, including Russia and North Korea, are breaking into systems with AI's help. He does not consider himself a doomer in the long run but expects a "dark age" as organizations work out how to adapt, and he noted the discussion also touched on managing AI-powered misinformation and slop on social networks.
More cybersecurity news from TechManNews.
