AI
Cognition Security Chief Says AI Is Pushing Cyberattacks Into a High-Pressure Phase
Alex Stamos, chief security officer at Cognition and lecturer at Stanford’s Cyber Policy Center, said in an interview that AI is improving both vulnerability discovery and exploitation capabilities, with private and state-backed attackers using it to carry out intrusions. He believes the cybersecurity industry will go through a “dark age” of adapting to the new technology.
This Decoder interview is the first part of a series on real-world AI applications. Stamos said October is traditionally Security Awareness Month, but major vulnerabilities such as those in PHP and Node have recently emerged, multiple companies and new AI-related cloud service providers have been breached in succession, and the Binance hack and the Cursor incident also show that attacks are being chained together.
He explained that one known chain so far is that after AI service provider Fireworks was compromised, API keys were stolen, allowing attackers to enter Cursor and then push malicious plugins, affecting 5,000 companies and creating 5,000 potential intrusions. AI helps attackers discover and exploit vulnerabilities at the same time, enabling state actors such as Russia and North Korea as well as private attackers to break into systems more broadly.
Stamos believes this state of affairs feels like a frontier similar to when the always-on internet world first emerged in the 1990s, with many weak entry points giving attackers opportunities while defenders are still scrambling to respond. He is not a long-term pessimist, but he judges that the industry will experience a “dark age” before it figures out how to adapt to the AI era. The interview also addressed the governance of AI-generated disinformation and low-quality content on social networks.