Strike Back Group · Against the motion

Chat Control

breaks the lock for everyone.

Open with an image: imagine the postal service opening every letter you send and reading it, just in case one of them is criminal. Nobody would accept that for paper letters. Chat Control does exactly this for every message, photo and video sent in the EU.

Say clearly: we are not against protecting children. Everyone in this room wants that. We are against a method that breaks security for 450 million people and does not even work well.

We all want to protect children.That is not the debate.

The debate is whether Europe should read the private messages of 450 million people to do it.

  • Security
  • Rights
  • Effectiveness

This is our whole case. Repeat these three words through the debate: security, rights, effectiveness. Judges remember structure.

Even if the other team wins one of them, they still have to beat the other two.

If they say "so you don't care about children": we propose better tools at the end, and the European Parliament itself, which wants to protect children, voted to keep encrypted chats out of scanning.

Every message.Every photo.Every link.Scanned.

450,000,000people who could be scanned. No suspicion needed.

"Chat Control" is the nickname for an EU law proposed in May 2022. The key tool is the detection order: an authority can order WhatsApp, Instagram or Gmail to scan the private content of its users.

It looks for three things: known abuse images (fingerprints), new images and "grooming" conversations. The last two need AI that guesses. Guessing means mistakes.

Encrypted apps were not excluded in the Commission's text, so the only technical way is client-side scanning: on your phone, before encryption.

If they say "only criminals are scanned": a detection order applies to a whole service, so every user of that service is scanned. That is why the Parliament wanted scanning limited to suspects with a judge's order.

Read before it's encrypted.

EU Centre
Police

500+ scientists from 34 countries call it technically infeasible.

The other side will say: "We're not breaking encryption, we're only scanning on the device." Answer: if a program reads your message before it is locked, the promise is already broken. It's a backdoor that sits inside your phone instead of on the server.

Apple quote (2023), from the company that actually built a scanning system in 2021 and then cancelled it: scanning "would create new threat vectors for data thieves to find and exploit".

The scientists' letter: over 500 cryptographers and security researchers from 34 countries call the plan "technically infeasible". "Bugs in Our Pockets" (2021, by Diffie, Rivest, Schneier and others): client-side scanning "does not guarantee efficacious crime prevention nor prevent surveillance".

Extra if needed: Apple pulled encrypted iCloud from the UK (Feb 2025) rather than build a backdoor; Signal said it would leave the EU rather than build "a surveillance machine".

Four years. Still no deal.

  1. A temporary law allows voluntary scanning.

  2. The Commission proposes mandatory scanning.

  3. Parliament: only suspects, never encrypted chats.

  4. Every compromise text fails.

  5. Council: “voluntary” scanning, made permanent.

  6. Parliament: 311 against, 228 for. Scanning lapses.

  7. Revived until 2028, encrypted chats excluded.

  8. Sixth round of talks: no deal.

  9. Still negotiated behind closed doors.

Don't read every date. The message: for four years, every time the EU tried to make scanning mandatory, it failed, because experts, lawyers and the Parliament kept saying it breaks fundamental rights.

Two strong moments: in March 2026 the Parliament refused even the voluntary extension (311 against, 228 for). When it was revived in July 2026 (Reg. 2026/1881), it had to explicitly exclude end-to-end encrypted chats. The EU's own temporary law now admits encryption must be left alone.

"Trilogue" = closed-door negotiation between Commission, Parliament and member states.

If they say "the Council already dropped mandatory scanning": the Council made "voluntary" scanning permanent and added risk-mitigation duties that critics say can force scanning through the back door.

A backdoor is a key under the doormat.

PoliceHackersHostile statesInsiders

A key doesn't check who is holding it.

A backdoor is "a typically covert method of bypassing normal authentication or encryption".

Use the doormat: the key is meant for the police, but a key doesn't check who is holding it. The famous 2015 MIT paper by top cryptographers is literally called "Keys Under Doormats": special access for governments creates new vulnerabilities everyone can exploit.

From our group document: once a backdoor is installed, whoever controls it can look into the system whenever they like. A huge privacy problem and a huge security problem at the same time.

It already happened.

100+

phones tapped through the network's built-in police wiretap, including the Prime Minister's.

Greece · 2004
1,000,000+

users' call records exposed. Chinese state hackers got in through the wiretap systems US law requires.

USA · 2024
1 yr

after launch, a flaw broke the US government's own backdoor chip. It was abandoned.

Clipper chip · 1994
“Encryption is your friend.”US cybersecurity agency (CISA), Dec 2024

Our strongest evidence: it is not a theory, it already happened. Twice, systems built so that "only the authorities" could listen were taken over by someone else.

Greece, the "Athens Affair": the legal wiretap function inside Vodafone's network was hijacked to spy on the Prime Minister and about a hundred top officials for months.

USA: CALEA (1994) forces phone companies to build wiretap access for police. In 2024 Chinese state hackers used that very access. Then the FBI and CISA told citizens to use encrypted apps like Signal and WhatsApp. The government that built the backdoor ended up recommending encryption.

If they say "the EU system will be secure": the US and Vodafone thought so too. A door only the good guys can open is like a gun that only shoots bad guys.

AI finds hidden doors faster than anyone can lock them.

27year-old bug in OpenBSD, found by Anthropic's Claude Mythos
81new Firefox flaws found by OpenAI's GPT-6 Cyber
948Linux kernel exploits built with its help

A scanner on 450 million phones would be the most valuable target on Earth.

From our document: AI can now break through protection layers. Very current: GPT-6 Cyber came out less than two weeks ago.

Security researchers used to find a few bugs a year by hand. These models find them by the thousands, including in code humans checked for decades: a 27-year-old OpenBSD bug, a 16-year-old FFmpeg bug that automated tests had hit 5 million times without catching.

The companies themselves say the models are so dangerous they only give them to verified security teams (Claude Mythos 5.1 only through Anthropic's Cyber Verification Program). And they leak: Mythos Preview was reportedly accessed by unauthorised users on the day it was announced.

Be precise if challenged: these numbers are reported by the companies; independent researchers have questioned some. The trend is not in doubt.

Privacy is not a setting. It's a right.

Art. 12

“No one shall be subjected to arbitrary interference with his privacy, family, home or correspondence.”

Universal Declaration · 1948
Art. 8

“Everyone has the right to respect for his private and family life, his home and his correspondence.”

European Convention · 1950
Art. 7

“Everyone has the right to respect for his or her private and family life, home and communications.”

EU Charter · 2000
Art. 15

“Freedom and confidentiality of correspondence and of every other form of communication are inviolable.”

Costituzione · 1948
Violated

Fundamental rights are not favours the state gives us; they are limits on what the state can do to us. Over 185 national constitutions mention the right to privacy.

Local touch: our own Constitution, Article 15, says the secrecy of correspondence is inviolable and can be limited only by a reasoned decision of a judge. Chat Control scans first, with no judge and no suspicion.

The EU's own test. Chat Control fails.

EU Charter, Art. 52: a right can be limited only if the limit is…

  • Provided by lawThat's what they are writing.Pass
  • NecessaryTargeted investigations already exist.Fail
  • ProportionateEveryone is scanned to find a few.Fail
  • Respects the essenceConfidential messages stop existing.Fail

Apply the EU's own test from Article 52 of the Charter: a limit on a right must be provided by law, necessary, proportionate, and must respect the essence of the right.

Chat Control fails the last three. Scanning everyone to find a few is not proportionate. And if every message can be read, confidential communication simply stops existing. A law that touches the essence of a right is invalid: no balancing can save it.

Not activists. The EU's own lawyers.

The basis for de facto generalized and indiscriminate scanning of private communications.

EU data-protection authorities · 2022

A “serious risk” of compromising the essence of the rights to privacy and data protection.

Council Legal Service · 2023

Forcing decryption weakens encryption for all users. A violation of Article 8.

European Court of Human Rights · 2024

Authority argument: we are not the only ones saying this.

The Council Legal Service phrase is very strong: "compromise the essence" of the right. Under the EU Charter, a law that touches the essence of a right is automatically invalid.

The Human Rights Court case: Russia ordered Telegram to decrypt messages. The Court said this would weaken encryption for all users and violated the right to private life.

Extra: the Parliament's own research service (EPRS, 2023) found no existing technology can detect abuse material without substantial error rates. And in 2014 the EU Court of Justice struck down the Data Retention Directive (Digital Rights Ireland, C-293/12) for general, indiscriminate surveillance. Chat Control repeats the mistake on message content.

Scan everyone. Accuse the innocent.

Only 1 in 5reports from platform scanning was real abuse material.

Ireland · 852 of 4,192 reports

3,000,000false alarms a day.

Our illustration · 1 billion images × 0.3% error rate (Ofcom test)

Each one is someone's private photo, seen by strangers.

When you scan billions of messages, even a tiny error rate produces a mountain of false alarms.

Ireland: only about 1 in 5 reports was real abuse material (852 of 4,192; 471 were plain false positives). Germany: the federal police say almost half of the US reports are not even criminal under German law.

The 3 million is our own illustration, say so clearly: a realistic volume times the error rate the UK regulator Ofcom measured for a PhotoDNA setting in testing. The point is the scale.

Second half: real offenders are the most motivated to avoid scanning. They switch to self-hosted or foreign apps, or encrypt files before sending. So ordinary people get scanned and criminals get away. The Parliament's researchers also warned that consensual images between teenagers can be flagged as abuse material.

TodayChild abuse.

TomorrowTerrorism.Copyright.“Fake news.”Protest plans.Your sources.Anything.

Whoever controls the list decides what gets flagged. Once built, it spreads.

"Mission creep": a tool built for one purpose slowly gets used for others. Changing the list is cheap and invisible to users.

Apple, 2023 (a company that had built the tool): "How can users be assured that a tool for one type of surveillance has not been reconfigured to surveil for other content such as political activity or religious persecution?"

Europol, the EU police agency, reportedly asked for access to all the scanning data, even false positives, with no limits (meeting minutes reported by Balkan Insight, 2023). Say "reportedly".

Other countries: every attempt caused damage or resistance. If the EU builds scanning into every phone, it gives every other government, including authoritarian ones, the perfect excuse: "Europe does it." Governments change; a tool built by today's democracy can be inherited by tomorrow's less democratic one.

Same idea. New names.

  • Detection orders2022
  • Upload moderation2024
  • “Voluntary” detection2025
Rules for you. Not for them.

A leaked 2024 Council draft exempted police, military and intelligence accounts from scanning.

From our document: the EU did not fully show how it wanted to implement Chat Control. Every time a version was rejected, it came back with a new name.

The most shocking detail: a leaked 2024 draft wanted to exempt the accounts of police, military and secret-service staff. If it's safe and harmless, why protect their own chats from it?

The Commission was reprimanded by the EU's data-protection supervisor for promoting this law with micro-targeted ads on X based on political opinions. Selling a privacy-invading law by invading privacy.

Also: the EU Ombudsman found maladministration in Europol letting an official move to Thorn, a company selling scanning AI (2025). Trilogues happen in private; most of what the public knows comes from leaks.

If a government orders a backdoor, is building it a betrayal?

“The equivalent of a master key, capable of opening hundreds of millions of locks.”
Tim Cook, Apple, refusing an FBI order · 2016

Legal is not the same as right.

The "traitor" point from our document, said carefully. If the architect builds the backdoor, who did they betray? The people who trusted the app.

Being ordered by a government doesn't remove responsibility. Engineers have professional ethics like doctors and lawyers: the ACM code says "Avoid harm" (1.2) and "Respect privacy" (1.6). And saying no is possible: Apple refused the FBI in 2016.

If they say "the law is the law": many laws have been struck down for violating fundamental rights. In 2014 the EU Court of Justice cancelled the EU's own data-retention law for exactly this reason.

Protect children. Without spying on everyone.

Targeted, not mass

Investigate suspects with a judge's warrant.

Remove it at the source

Take abuse material down where it is published.

Safety by design

Private-by-default for minors, easy reporting.

People, not scanners

More investigators, support for victims, education.

Very important for the judges: we don't just say no, we propose. Being against Chat Control is not being against child protection.

Targeted investigations are the European Parliament's official position, so it's not an extreme idea. Removing material at the source was agreed in principle in September 2026.

The false alarms from mass scanning actually waste police time that could go to real cases.

A lock that opens for everyone is not a lock.

Vote against Chat Control

  • Breaks security
  • Violates rights
  • Doesn't work

Strike Back Group · Thank you

Slow down. Repeat the three words: security, rights, effectiveness.

Closing line: "The other side asks you to trust that a key under the doormat will only ever be used by the right people. History says otherwise: Greece, the United States, and now AI that finds every hidden door. Protect children with targeted, effective tools, and keep the lock on everyone else's door. Vote against Chat Control."

Sources

  1. European Commission, Proposal for a Regulation laying down rules to prevent and combat child sexual abuse, COM(2022) 209EU document
  2. Regulation to Prevent and Combat Child Sexual AbuseWikipedia
  3. Council of the EU, Council reaches position on law protecting children from online abuse, 26 Nov 2025EU press release
  4. eucrim (Max Planck Institute), Interim rules on voluntary CSAM detection reinstated until 2028, Aug 2026Legal journal
  5. P. Breyer, Chat Control 2.0 trilogue update, 30 Sep 2026Former MEP, campaigner
  6. EDPB-EDPS Joint Opinion 04/2022, 28 Jul 2022EU authority
  7. European Parliamentary Research Service, complementary impact assessment, Apr 2023EU study
  8. ECtHR, Podchasov v. Russia, no. 33696/19, 13 Feb 2024Court judgment
  9. EFF, European Court of Human Rights confirms undermining encryption violates fundamental rights, Mar 2024NGO analysis
  10. Human Rights Research Center, Child protection or mass surveillance?, Sep 2026Research centre
  11. Backdoor (computing)Wikipedia
  12. End-to-end encryptionWikipedia
  13. Abelson et al., Keys Under Doormats, MIT, 2015Academic paper
  14. Abelson et al., Bugs in Our Pockets: The Risks of Client-Side Scanning, 2021Academic paper
  15. Scientists' open letter on the CSA Regulation, Sep 2025Open letter
  16. 2004–05 Greek wiretapping caseWikipedia
  17. Communications Assistance for Law Enforcement ActWikipedia
  18. Salt TyphoonWikipedia
  19. NBC News, US officials urge Americans to use encrypted apps, Dec 2024News
  20. Clipper chipWikipedia
  21. MacRumors, Apple explains why it abandoned CSAM detection, 1 Sep 2023News
  22. MacRumors, Apple pulls encrypted iCloud feature in UK, 21 Feb 2025News
  23. The Record, Signal calls on Germany to vote no on Chat Control, Oct 2025News
  24. Apple, A Message to Our Customers, 16 Feb 2016Company statement
  25. ACM Code of Ethics and Professional Conduct, 2018Professional code
  26. Fundamental rightsWikipedia
  27. Right to privacyWikipedia
  28. Universal Declaration of Human Rights, 1948UN
  29. European Convention on Human RightsCouncil of Europe
  30. Charter of Fundamental Rights of the European UnionEU law
  31. Costituzione della Repubblica Italiana, art. 15Senato della Repubblica
  32. CJEU, Digital Rights Ireland, C-293/12, 8 Apr 2014Court judgment
  33. ICCL, An Garda Síochána retains files on innocent peopleNGO, police data
  34. netzpolitik.org, NCMEC reports and BKA figuresNews
  35. Ofcom, Illegal Harms Statement, vol. 2, Dec 2024UK regulator
  36. EU Reporter, Leak: interior ministers want to exempt themselves, Apr 2024News
  37. heise online, EDPS rebukes Commission over chat control ads, Dec 2024News
  38. noyb, Complaint against the Commission over targeted chat control adsNGO
  39. Techzine, Europol sought unlimited access to data from chat services, 2023News
  40. netzpolitik.org, Ombudsman criticises Europol–Thorn revolving door, 2025News
  41. Assistance and Access Act 2018 (Australia)Wikipedia
  42. India IT Rules 2021Wikipedia
  43. Anthropic, Project Glasswing, Apr 2026Company announcement
  44. Claude MythosWikipedia
  45. Claude Platform docs, Claude Mythos 5.1Documentation
  46. The Register, OpenAI releases GPT-6 Cyber, 5 Nov 2026News
  47. SecurityWeek, OpenAI says GPT-6 Cyber hunts flaws without crossing the line, Nov 2026News
  48. European Parliament, LIBE report A9-0364/2023EU document