AI could make software too secure, leading to law enforcement 'going dark'

The Paradox of AI and Security

The author, following attendance at Usenix Security, raises a concern regarding the future impact of AI on software security. The central worry is that AI will make software excessively secure, creating an unintended consequence for law enforcement and intelligence agencies.

Law Enforcement 'Going Dark'

This heightened security, driven by AI, is predicted to cause U.S. intelligence and law enforcement agencies to 'go dark'. This term signifies a substantial loss in their ability to conduct surveillance and access digital information, which has implications beyond just the agencies themselves, affecting general computer security and privacy.

Historical Context of Surveillance Challenges

To illustrate the potential shift, the author references the evolution of electronic surveillance. The early 2000s, as depicted in 'The Wire', involved wiretapping payphones and burner phones. This scenario became outdated with the rise of smartphones and encryption technologies.

The late 2000s saw smartphones become a new source of data for law enforcement. However, around 2010, Apple began encrypting iPhone data, followed by end-to-end encryption in iMessage. By 2016, WhatsApp had nearly a billion users utilizing default end-to-end encrypted messaging, demonstrating a rapid shift in digital communication security.