OpenAI starts labeling text from ChatGPT in the EU

OpenAI is now launching text labeling in the EU. The idea is that text created with OpenAI's tools should be recognizable as AI-generated. The background is the EU's AI Act, which requires transparency around AI content.

What does it mean for you personally? If you use AI for a greeting, an application or an explanation, there is hardly anything you need to do differently. But if you pass text on without reading it through, a label could in principle reveal that AI had a hand in it.

What does it mean for your company? If you use AI for newsletters, website copy or customer emails, it is wise to check how the labeling works in practice and whether it follows the text when it is copied into other systems. It is also a good occasion to have a simple internal agreement on when AI use should be disclosed openly.

Rule of thumb: If you are an ordinary user, you can safely wait and see how it will work. If you publish text on a large scale as a company, set aside a few minutes in the coming weeks to read OpenAI's own description and talk to whoever is responsible for your content.

What it means for you

If you use ChatGPT for ordinary things like a greeting or an application, you do not need to do anything differently, but you cannot count on AI use staying hidden if you pass the text on without reading it through. If you work with newsletters, website copy or customer emails, set aside a few minutes to read OpenAI's own description of the labeling. Also talk to whoever is responsible for your content, and briefly agree on when you will say openly that AI has helped.