Encryption backdoors: The trade-off between security and privacy

Encryption backdoors: The trade-off between security and privacy

Cryptographic backdoor controversy in ProtectEU strategy

The European Commission has put "legal access" on the table again in ProtectEU. This is not a new script-from the Clipper Chip in 1993 to the Investigative Powers Act in 2015 to the European Union's Child Sexual Abuse Regulation (CSAR) in 2022, end-to-end encryption (E2EE) is being reinvented every few years as a "safe haven for criminals." The difference is that this time, under the guise of "protecting the internal security of the EU," law enforcement, intelligence, border control, and customs protection are packaged into a "security package" in an attempt to solve mathematical problems with administrative orders. The real question is not whether technology can do it, but the efficiency-oriented logic that defines "security" as "the country can see everything at any time" and regards privacy as a quantifiable, tradable, and scalable cost item.

Mathematics does not talk about politics. The security model of end-to-end encryption is based on the premise that "only the two parties holding the private key can decrypt"-once backdoor mechanisms such as "master key","fragmented escrow" and "client scanning" are introduced, the attack surface will change from "cracking a single session" has expanded to "breaking a centralized key infrastructure." The 2015 Keys Under Doormats report (signed by 14 cryptographers including Diffie, Schneier, and Rivest) demonstrated that any access mechanism reserved for law enforcement is essentially a structural breach of the integrity of the protocol. The CSAM client scan that Apple wanted to promote in 2021 was pointed out by the security community that it could be redirected to any content review tool, and was eventually forced to withdraw it. Signal President Meredith Whittaker bluntly said: If the EU forces backdoors, Signal will withdraw from the EU market. This is not a threat, it is an engineering reality: a code base without backdoors cannot maintain two versions.

Efficiencyism is best at compressing systemic risks into controllable KPIs. For policy makers,"encryption hinders solving cases" is a clear pain point, and the crime detection rate, prosecution rate, and conviction rate are all hard indicators; and the second-order consequences of "the chilling effect of national communications being monitored","exposure of journalists 'sources","The channels for help for victims of domestic violence have been cut off" are difficult to quantify, lag behind, and no one is held accountable. This is like a family tearing down the bedroom door for "efficiency", pasting the safe password on the refrigerator, and giving each relative a master key-the reason is that "there are no thieves at home, so it is convenient to check things." In the first three years, time was indeed saved to find the key, until one day the nanny boyfriend copied the master key and took away the family's savings. After the review, it was discovered that the hundreds of hours of "opening time" saved were exchanged for irreversible asset clearing. The trap of efficiency is that it is only the numerator of explicit benefits, not the denominator of implicit risks.

There is a Berlin startup founded by a couple who supports eight employees by making encrypted note apps. Last year, the German Federal Criminal Police Office (BKA) came to the door and requested that a decryption interface that was "judicial authorization only" be implanted in the service, promising to "never abuse" and "only for serious crimes." The founders refused for the simple reason: their threat model never assumed "trustworthiness on the server." Once the interface is implanted, databases, logs, backups, CI/CD pipelines, and even notebooks of retired employees become attack surfaces. BKA replied: "That's your engineering issue. Legal compliance is the bottom line." The couple eventually closed the company and moved to Switzerland with the code base. Eight engineers are unemployed, Europe is missing a privacy tool, and BKA still doesn't have a back door. This is not an exception- Threema, Wire, Proton are all evaluating evacuation plans. Efficiency-based accounting is never a "trust withdrawal cost."

Must steelman's claim on the enforcement side: Child sexual abuse material (CSAM) does proliferate on crypto platforms. The European Union received 1.5 million NCMEC clues in 2023, a year-on-year +23%; German BKA statistics show that encrypted communications caused about 40% of felony clues to be interrupted. These figures are not made up. They are real for the children victims and the investigation is blocked. But equating "unbreakable encryption" with "unbreakable crime" is a logical leap. Europol's own 2022 report acknowledges that traditional methods such as metadata analysis, device-side forensics, penetration testing, social workers, informant networks, and international cooperation still solve more than 85% of cryptography-related cases. What the backdoor solves is only the part of the cost of laziness that "does not want to spend energy, wait for authorization, and collaborate across borders." Using national privacy as fuel burns laziness in law enforcement capacity building.

Business accounts are clearer. Who is lobbying? Look at the EU Transparency Register in 2024: Cellebrite, MSAB, Grayshift, Magnet Forensic-these mobile phone forensics giants have annual revenue exceeding 300 million euros, and their lobbying budgets will only increase but not decrease. The essence of their business model is to "sell the key to the government to open mobile phones." Once the law enforces the back door, their products will be legalized from an "attack tool" to a "compliance interface" and the valuation will double. At the same time, companies such as Signal, Threema, Element, and Proton that do not sell user data have no lobbying budget and only have the "exit the market" card. This is not a technology debate, it is rent-seeking: legislation is used to define "security" as "products from specific suppliers must be purchased."

The second sequence of consequences is the most subtle and fatal. In 2022, a Polish journalist was monitored by Pegasus for reporting on government corruption. His source list was leaked after encryption applications were forcibly implanted into a "compliance interface"-it was not stolen by hackers, but the "legal access" rights were abused. There are precedents in Mexico, India, and Hungary: anti-terrorism law tools are ultimately used to monitor opposition, journalists, and judges. Article 14 of the ProtectEU text allows "national security" exceptions to judicial authorization, which is a blank check in the hands of the current governments of Hungary, Poland, and Slovakia. People who say "I have nothing to hide" often ignore that privacy is not for "bad people" but for people "defined as bad people by power." When the definition of "security" is in the hands of the ruling party, encryption backdoors are political weapons that can be activated at any time.

The price of efficiency has never been the number on the explicit bill, but the moment when the system loses its resilience. Eliminating encryption as a "friction that hinders solving cases" is like removing the immune system as an "energy-consuming burden"-it does save energy in the short term, and you will die completely when encountering a virus. The "security" that the European Commission wants is to run the communications of 450 million people naked in front of countries, suppliers, and an unknown future attacker in exchange for a possible 3-5 percentage point increase in the detection rate. This account is calculated accurately and ruthlessly, but it is not long-term. The real security is not "no one can hide secrets", but "no one can look at secrets at will." The next time you see the headline "Privacy must be compromised for security", ask: Who received the commission in this deal? Who bears the final risk?