The cost of confusing data volume with intelligence
Once again, the EU is trying to push a bill that would allow scanning private communications for illegal content — all in the name of fighting terrorism and, of course, the golden ticket: child abuse.
Don’t get me wrong, child abuse is absolutely horrific. But this bill is not the solution.
Luckily, the bill was removed from the EU’s agenda — but this axe is still hanging over all of us.
Not the solution
This bill simply doesn’t work — and it will backfire in ways few people seem to understand. I already wrote about this before, in this article.
Just imagine the EU wanting to scan all of our private traffic. Let’s put that into perspective — every message, every attachment, every emoji being scanned.
Here are some numbers to frame that idea:
- Whatsapp: 100 billion messages daily
- Signal: 2 billion messages daily ( assumption, no data found )
- Telegram: 20 billion messages daily
- Email: 376 billion messages daily
- SMS/Text messages: 25 billion messages daily
Over 500 billion messages every single day. And let’s assume each message — including emails — averages around 2 KB in size.
That’s over one petabyte of data per day.
To put that in perspective: that’s about 1,000 TB, or roughly 50 hard drives of 20 TB — filled every day.
How are we supposed to scan 500 billion messages every single day? You’d need some serious hardware just to process that. And of course, they’ll probably want to throw some AI or LLMs into the mix — which means even more hardware.
Then there’s data retention, backups, redundancy, monitoring… the list goes on.
I asked ChatGPT to come up with a rough estimate of what this kind of infrastructure would actually cost.
| Component | Unit / Basis | Rough estimate |
| Compute (≈12,000 vCPUs) | Cloud or on-prem | ~€5–10 million per year |
| GPU cluster (≈10,000 × NVIDIA A100/H100) | ~€20 000 each | ~€200 million CAPEX |
| Storage (≈60 PB usable) | ~€25 per TB + infra | ~€1.5 million CAPEX |
| Networking, power, cooling, ops & security | — | ~€20–30 million per year |
| Total | €250–300 million in upfront cost, plus €40–50 million yearly operations |
False positives
And then there’s the issue of false positives.
I send a perfectly innocent message that somehow gets flagged — what happens then?
Someone will have to review it. Which means… more skilled investigators.
And what would the false positive rate even be?
Even if it’s as low as 0.0001%, that still means 500,000 messages flagged per day — all false alarms.
That is unworkable. Full stop.
Police force
Now imagine that a criminal investigator in the Netherlands earns around €70,000 a year. That means that with €50 million in running costs, you could hire over 700 highly skilled investigators.
Not to mention, you’d still need investigators to review all those false positives anyway.
So, which is more productive?
Spending €300 million on computer hardware — plus the same investigators chasing false alarms — or hiring 700 professionals who actually investigate the real thing?
Plain sight
Then there’s the simple option: hide your messages in plain sight.
Take the 1st letter of the 1st word, the 2nd letter of the 2nd word, the 3rd of the 3rd, then the 2nd of the 4th, and so on — a tiny letter-position trick that turns an ordinary sentence into a courier pigeon. With tools like ChatGPT, generating perfectly natural-sounding cover text that encodes a short secret is ridiculously easy.
It doesn’t even matter how you encode it; even something as trivial as ROT13 would likely slip past any AI or LLM.
Bad actors
Let me ask you a simple question — and I’ve asked it before:
Do we really think bad actors would keep using all these backdoored platforms?
Hell no. They’d move to other channels, use real encryption, and vanish off the grid.
So who is this bill actually aimed at?
Yes.
Us. Civilians. Journalists. Your kids. My kids.
Me.
You.
And definitely not the hardened criminals, terrorists, or child abusers of this world.
Math
And the most simple fact: encryption is based on math — almost basic math, I’d say.
So… are we going to ban math now?
Conclusion
This bill is a stupid idea — a can of worms we really don’t want to open. It won’t protect us from terrorists, and it certainly won’t protect our children from predators.
Instead, it will hurt all of us, not the criminals it claims to target.
You and I will become the suspects when we accidentally end up in that pile of false positives.
Good luck.
Oenva bhg (ROT13)