In August 2026, cryptography professor Matthew Green sparked debate within the cybersecurity community by suggesting that advancements in AI could drastically reduce software vulnerabilities, potentially limiting governments' ability to use hacking tools for lawful surveillance. Green expressed concern that AI-driven bug detection might lead to stronger software security, eroding the current “uneasy truce” where governments avoid backdoors in favor of exploiting security flaws. Historically, law enforcement agencies have grappled with encryption technologies, such as end-to-end encryption on messaging platforms, which hinder traditional surveillance methods, but have compensated by purchasing sophisticated spyware and hacking tools.

Experts in cybersecurity and offensive security firms weigh in on Green's theory with mixed views. Some agree that there is currently a “gold rush” of bugs due to AI’s ability to uncover them rapidly, but predict this phenomenon is temporary as bugs will eventually become scarce. Others argue that while easier bugs may disappear, complex vulnerabilities valuable for government surveillance will persist and that AI can assist researchers in finding these. Additionally, some professionals highlight that modern security measures in devices pose a bigger challenge to hacking efforts than AI’s influence on vulnerability availability.

The ongoing dynamic has significant implications for debates around device security and privacy, especially regarding calls to introduce backdoors for government access. While AI may make software more secure overall, some experts warn that governments might renew pressure for built-in access points due to the diminishing returns from traditional hacking tools. Privacy advocates caution that authoritarian regimes, in particular, will continue seeking “exceptional access” to communications, potentially threatening overall security.

Industry insiders, including Katie Moussouris of Luta Security, anticipate that the transition to fewer exploitable bugs will not be immediate and estimate that substantial impacts on intelligence agencies’ hacking capabilities might emerge after the next U.S. presidential election. Until then, the balance between privacy, security, and law enforcement access remains precarious as AI reshapes the cybersecurity landscape and could redefine how governments pursue digital surveillance.