The Crypto Wars Ended, And The Prize Went To Whoever Owns Your Endpoint

Cryptography engineer Matthew Green's essay "Going Dark, and the era of law enforcement hacking" reached the Hacker News front page with 179 points and 109 comments ^[1]. The argument, as relayed in a dev.to summary of it, is that encryption won the wiretap debate and that the win simply moved the fight to the endpoint, which is the one layer most organisations manage worst ^[2][3].

For decades agencies framed end-to-end encryption as a "going dark" problem and asked for exceptional access: warranted decryption built into the product, pushed by the FBI, DOJ and intelligence agencies elsewhere ^[4]. The cryptography community's answer was close to unanimous, and it has not changed: a weakness reserved for the good guys is not a buildable artefact, because criminals and foreign services eventually use the same door ^[5]. Green's claim, per the summary, is that law enforcement quietly stopped fighting that battle because it found a substitute ^[6].

The substitute is mechanical rather than clever. Traffic is encrypted in transit and decrypted on the handset, so the handset is the weak link ^[7]. The summary lists four properties that make this attractive to an agency: it works against any encryption without standards fights or vendor negotiation ^[8]; it is targeted per device rather than population-wide ^[9]; it is warrant-based and, in that view, increasingly treated as constitutional ^[10]; and it is effective, with modern spyware able to compromise fully patched devices ^[11].

There is a supply chain behind that. NSO Group in Israel makes Pegasus, which can remotely compromise iOS and Android ^[12]; Cytrox in North Macedonia makes Predator, used by governments worldwide ^[13]; Intellexa in Greece is described as an alliance of surveillance companies ^[14]; FinFisher in Germany was among the earliest ^[15]. That is four vendors across four national jurisdictions, which is why export control is a slow instrument here ^[16]. The Pegasus Project investigation found NSO clients using Pegasus against journalists, activists and political opponents in dozens of countries ^[17].

The asymmetry that matters to operators is legal, not technical. US wiretapping requires a Title III warrant with strict minimisation procedures ^[18], while device hacking often proceeds under Rule 41 of the Federal Rules of Criminal Procedure, which carries fewer safeguards ^[19]. There is no statutory framework specific to government hacking ^[20], no requirement to disclose vulnerabilities discovered in the course of it, and no clear rule on what happens to data taken off a compromised device ^[21]. No amount of protocol review fixes that.

The market prices the shift plainly enough. Zero-day brokers pay $2 to $3 million for iOS zero-click exploits ^[22], alongside government acquisition programmes such as the US vulnerability equities process ^[23] and AI-assisted vulnerability discovery ^[24]. The defensive column is entirely endpoint work: iOS Lockdown Mode, Android's enhanced sandboxing, Secure Enclave and Titan M, memory-safe languages in kernel code, and mitigations including PAC, BTI and MTE ^[25]. Not one of those is a wire-level control ^[26].

One tension in the source is worth naming, because it decides budgets. The same post says spyware compromises fully patched devices ^[11] and that most spyware exploits known vulnerabilities, making patching the first line of defence ^[27]. Both can hold across a tiered market, but they imply different spending: patch latency and Lockdown Mode for the broad case, and a different order of paranoia for staff who are individually worth a seven-figure exploit.

Worth watching: whether any statutory framework for government hacking appears, whether zero-click prices move as mitigations land, and whether exploit provenance starts surfacing in discovery.