Watermarking AI Text in the EU Is a Toothless Fix (For Now)

A 92 percent detection rate sounds great until someone edits the sentence.

OpenAI will watermark ChatGPT text by default in the EU with textGrain, hitting 92 percent detection in tests.

OpenAI itself notes the signal collapses the moment you edit the output, which is exactly what humans do.

So the watermark catches the lazy copy-paste and misses the deliberate paraphrase. That's most of the risk.

Outside the EU it stays off, so the 'global' fix is really a regional band-aid.

I'm glad it exists. I just wouldn't call it a solution to disclosure.

Put one of these tools on the leaderboard

Every rank on Best AI Utility is paid and time-stamped — outbid the current holder with one crypto payment and the rank is yours until someone outbids you.

ChatGPT is #1 on the chatbots leaderboard

Current bid $1,500 · outbid from $1,505

Outbid now

Powiązane artykuły

Best AI Tools Directory in 2026: Find the Right AI Tool for Every Task

Too Many AI Tools? Here’s a Better Way to Find What You Actually Need

The Week AI Got Practical: Cyber Models, Cheap Video, and a Finance Claude

If you only check AI news on weekends, this was the weekend to check. The theme was not a bigger model, it was models that do a specific job well enough to hand them the keys.

GPT-6 Cyber Is a Bug Hunter That Stays in the Lines. That Changes Who Gets One.

A model that finds vulnerabilities but refuses to step outside authorized scope is not just a better scanner. It is the first time the powerful exploit-finder has been gated to the people who already had permission to look.

Odkryj więcej

Przewodniki Wszystkie narzędzia Porównaj narzędzia