SocietyAI glossary
Watermarking
An invisible watermark embedded in AI-generated content (images, video, audio) to prove its artificial origin, even after cropping or compression. Google's SynthID is the most widely deployed example, and the European AI Act makes this type of transparency mandatory from August 2026.
In other words
The watermark on banknotes, image and video version: invisible to the naked eye, detectable by machine.
Alyriaz articles that mention it
- Can a model's intelligence be stolen?
Model extraction involves reconstructing a system's capabilities by querying it at scale. We explain the mechanism, the possible defences, and why none of them really works.
- An AI text detector reported as bypassed: the race is lost, confirmed
A company claims to have developed a tool that neutralises the most accurate AI-text detectors. We look at why this race is structurally unwinnable and what that means.
- ChatGPT Images 2.5: three billion images a week, and a shift towards editing
OpenAI has released ChatGPT Images 2.5 with two models for developers, halved latency, and gains focused on editing. We analyse this shift and what it targets.
- How can you tell if a text was written by an AI?
Between the myths about em dashes and unreliable detectors, here's a look at what actually helps spot generated text, and why caution is still warranted.
- Claude's watermark bypassed in four hours
Anthropic has announced a watermark for texts produced by Claude. Code to remove it appeared four hours later. We explain why watermarking text is far more difficult than watermarking an image.
- What is C2PA? The invisible passport that will accompany every image
The C2PA is the technical standard that makes it possible to attach a verifiable history of origin to an image, video, or audio file. We explain how it works, why it upends the logic of detection, and what it cannot do.