Everything is about to “go dark”
Summary
Matthew Green argues that AI-driven vulnerability scanning will make software so secure that law enforcement and intelligence agencies will lose their ability to exploit devices, forcing a renewed push for intentional backdoors.
AI accelerates software hardening.
Defenders are integrating AI-based scanning into CI toolchains to patch decades of bugs, likely hitting a ceiling on remotely-exploitable vulnerabilities within two years.
major pieces of software are likely to run out of remotely-exploitable bugs.
Vulnerability monopolies are failing.
Despite US export restrictions on models like Anthropic's Mythos, Chinese labs like Z.ai and Moonshot demonstrate that vulnerability finding is not a single-lab monopoly.
vulnerability finding isn’t something that a single lab is likely to hold a monopoly on.
Scarcity drives backdoor demand.
As low-hanging bug fruit disappears, agencies will exert enormous pressure on industry to re-architect systems for intentional, exceptional access.
The demand for constructed, intentional backdoors will re-start in earnest.
Backdoors cause self-sabotage.
Mandated access mechanisms may only weaken US systems, leaving domestic infrastructure vulnerable to foreign adversaries while non-US governments abandon US software entirely.
This deliberate self-sabotage will happen just at a moment when we’re finally getting a handle on securing our own infrastructure.
Highlights
Just realizing this doesn’t mean that I have any strategy in mind to avoid it
in the real world, it does feel likely that we’re going to hit some sort of a ceiling on the number of useful bugs
Read the original ↗Save to your Sponge
Summarized with Sponge. Quotes are the author’s own words.