Going Dark: The Era of Law Enforcement Hacking
How AI is enhancing software security while impacting law enforcement capabilities.
1AI and Software Security
AI is significantly enhancing software security. However, this could severely limit law enforcement's electronic surveillance capabilities. Particularly in the U.S., intelligence and law enforcement agencies might enter a period known as 'going dark.' This situation arises as software becomes much more secure.
2The Evolution of Electronic Surveillance
In the early 2000s, electronic surveillance was primarily conducted through landlines and mobile phones. However, the proliferation of smartphones changed this landscape. In 2010, Apple began encrypting iPhone storage, making it harder for law enforcement to access phones. The use of end-to-end encryption by apps like WhatsApp accelerated this process.
3The FBI vs. Apple Standoff
In 2016, the FBI requested access to a locked iPhone following a terrorist attack. Apple refused the request. However, an external company announced it could hack the phone. This incident highlighted law enforcement's efforts to access encrypted devices. Tools like GrayKey and Pegasus made such access possible.
4AI and the Discovery of Security Vulnerabilities
Anthropic's Mythos model was optimized for finding software vulnerabilities. The U.S. government temporarily blocked its export. However, OpenAI and Chinese labs developed similar models. These models quickly detect and help patch security vulnerabilities in software.
5Progress on the Defense Side
AI enables the defense side to quickly patch security vulnerabilities. CI toolchains are being reconfigured to scan for vulnerabilities before software reaches human hands. This significantly reduces the number of remotely exploitable vulnerabilities. However, it complicates the work of law enforcement.
6The Risk of Law Enforcement 'Going Dark'
As software becomes increasingly secure, law enforcement's electronic surveillance capabilities could be severely limited. This could happen for the first time since 2010. Law enforcement agencies might face the risk of completely 'going dark' on advanced devices and software.
7The Future of AI and Security Vulnerabilities
AI is rapidly detecting and patching security vulnerabilities in software. This significantly reduces the number of remotely exploitable vulnerabilities. However, it complicates the work of law enforcement. In the next two years, major software might run out of remotely exploitable vulnerabilities.
8Software Security and the Future of Law Enforcement
AI is enhancing software security while limiting law enforcement's electronic surveillance capabilities. This increases the risk of law enforcement 'going dark.' However, it is seen as a positive development for software security. The future balance of this dynamic remains to be seen.