AI Models Are the New Bug Bounty Hunters
AI is changing the bug hunting game. Cisco's using AI to scan billions of code lines in weeks. Anthropic's expanding its Mythos program, but there's a catch.
JUST IN: The AI bug hunting scene is heating up, and Anthropic and OpenAI are leading the charge. They're turning what used to be a painstaking grind into a swift operation.
Cisco's AI Leap
Cisco's latest move? They've roped in AI to turbocharge their security checks. Anthony Grieco, Cisco's SVP, revealed their systems scanned 1.8 billion lines of code in just eight weeks. That's a job that would have taken human teams eight years. Wild, right?
But here's the kicker: Cisco hasn't said how many bugs they actually caught. Or if they've all been patched. Are they hiding something, or is it just an oversight?
Anthropic's Ambitious Expansion
Meanwhile, Anthropic is expanding its Project Glasswing. Initially exclusive, the program's now welcoming around 200 partners. That's 150 more than when they started in April.
These aren't just any partners. We're talking about powerhouses like Cisco, and now new players like Korea's KISA, Samsung, and Rubrik. Anthropic's Mythos is so potent at sniffing out vulnerabilities, they feared chaos if it fell into the wrong hands. And they might just be right.
The Bigger Picture
Here's where it gets interesting. Palo Alto Networks, another big name in cyber, used these AI models and uncovered 26 CVEs in a month. Normally, they'd find fewer than five. That's a massive leap.
With AI-powered exploits looming, organizations have a tight three-to-five-month window to get ahead. Will they make it, or is the gap closing too fast?
The labs are scrambling, and for good reason. As AI models become more adept at both finding and fixing bugs, the balance of power in cybersecurity is shifting. The question is, who's ready for it?
Get AI news in your inbox
Daily digest of what matters in AI.