European Threat To End-To-End Encryption Would Invade Phones

European lawmakers have been implementing a way to circumvent end-to-end encryption to address child sexual abuse material ("CSAM") - what some activists term the "Chat Control" law. End-to-end communication guarantees that if you communicate with someone, only a receiving device will be available, and the sending device can decrypt or see the message in question. This is close to Bitcoin's principles for spending money and financial flows - allowing you to choose who you reveal your spending to or who not to - with privacy-preserving and self-custody technologies also under attack. Freedom technology like Bitcoin and Nostr (which has recently built a proposal for end-to-end encrypted messages) and open-source tools like Signal focus on giving people digital privacy and autonomy - principles under attack from a recent proposal to scan users' devices for content before it is encrypted.

Experts in cryptography and other scientists/researchers have released an open letter to discuss this proposal's technical and policy issues. The urgency of the moment is because the legislation has a realistic chance of passing this time and, to the surprise of all, is being proposed right before the European elections. After several other attempts had been thwarted - as Bart Preneel, one of the originators of the letter explains, the Belgian presidency has advanced a new version of the Chat Control law, tweaked in two places to get more political support - the first tweak is that different services will be categorized into different levels of risk, with only "High Risk" services being scanned. However, the definition of a "High-risk" service is any that offers end-to-end encryption and real-time communications - apps like Signal and WhatsApp, used widely by so many, would fall under that definition. Services that allow for anonymous IDs (the default on the Internet) would also fall under this category.

The second tweak was made to address the issue of false positives. The proposal suggests triggering reporting only after a certain threshold is met - two for known CSAM and three for 'new' CSAM. However, this approach could lead to innocent individuals being flagged as suspicious. For instance, a parent who takes photos of their own child might do so multiple times, potentially triggering the algorithm and leading to false accusations. The sheer number of false positives, which could reach millions in a week, could make the proposed solution untenable and subject many innocent people to unwarranted suspicion.

While European courts might strike even this new version, as they've indicated that weakening encryption of secure communications is banned, it might still leave a lag time of a few years before the courts rule against the legislation. By this time, it might have morphed into another version that can be "acceptable." This is happening with data retention of electronic communications by European police services. While the courts have set their opinions, it can take time to rein back what legislators set in motion once rolling.

While Chat Control is presented as a way to allow for end-to-end encryption while also allowing for scanning content for authorities, the cure is much worse than the disease - and a "play of words," according to Praneel. Instead of a backdoor, an algorithm is proposed to be planted on every device before content is encrypted. This is "more intrusive [than a backdoor to end-to-end encryption], but officially, they can say we don't break the encryption; you still have your end-to-end encryption, so that's how they tried to sell it to the general public." But make no mistake - screening all user devices is more intrusive and harmful than even a backdoor to end-to-end encryption.

A simple thought experiment helps us delve into the potential consequences. This algorithm placed on devices is technically flexible: today, it's being used to scan for child abuse material and grooming - but what's to stop it from being deployed to scan for any amount of expression or content deemed unfavorable for the state?

Even if one trusts the European Union and its member states, it's easy to see how techno-authoritarian states will build on this technology for their people or use the European example to justify their repressive processes. In China, you can get arrested for using specific applications such as VPNs, or privacy-protecting options like Whatsapp, and using machine learning to censor and monitor communications is routine. It is disheartening that the European Union wants to follow parts of this path.

The scanning algorithm is also technically inelegant. There is the highlighted potential for false positives, which can flag innocent people at very high rates and make enforcement tricky. There is also a dependency on age verification at the source, of which there is no reliable technical standard. As Praneel also points out, there is no way to have a perceptual hash function that garbles the video/content in question and keeps recording it "legal" without publishing it publicly - Apple's attempt to implement screening had elegant cryptography in other places but had to release the hashing function - which allows people to evade the system if they research it sufficiently. Previous efforts to scan content on a server for unencrypted messages (ex, Facebook) are nowhere near having a scanning algorithm on user phones - similar in dynamic to changing banking from a relationship between a private entity and an individual that can choose cash to one that directly routes central bank digital currency to an individual and implements governmental scanning and control on individual people.

Lastly, there are alternatives available to reduce child abuse, which should be the primary goal, without sacrificing digital privacy for so many people - for example, more proactive reporting of abuse on social media platforms and legal tools to chase criminals that already exist - that don't involve scanning every device.

"The EU Commission's proposal pays a lot of lip service to the importance of privacy, but ultimately, it green lights the ability for national legislatures to clamp down on encrypted messaging. If authorities suspect a crime, they have every legal tool available to pursue suspects. EU citizens do not need to lose access to cryptographic protections just because police want their jobs done for them." said Yael Ossowski, deputy director of the Consumer Choice Center, an international consumer advocacy group, summing up the sentiment that European legislators can do much better at the balance between Internet privacy and national security. Luckily, some legislators like Patrick Breyer have warned against this new law, and it isn't a sure thing that it'll pass - but for people concerned about their privacy in a digital age, it would be a disaster if it did.