Cryptography professor Matthew Green has sparked debate in the cybersecurity community by suggesting that AI could make software so secure that governments lose their ability to lawfully hack suspects. His argument is that as AI-powered tools become better at finding and patching software vulnerabilities, the "uneasy truce" under which law enforcement uses hacking tools and spyware instead of demanding encryption backdoors could break down, potentially prompting governments to push for backdoors once again.

Green's warning follows years of "going dark" debates, dating back to 2014 when then-FBI director James Comey said encryption in apps like Signal, WhatsApp and iMessage was hampering investigations. Since then, agencies have relied on purchased exploits and spyware rather than backdoors to access devices. TechCrunch canvassed several security professionals, including researcher Luna Tong and Crowdfense CTO Paolo Stagno, finding mixed views: some agree AI-driven bug-hunting is causing a temporary "gold rush" that will make flaws scarcer, while others doubt states would ever abandon surveillance capabilities regardless of how AI reshapes the vulnerability landscape.