Office management edition
ChatGPT can now show answers with buttons and charts. New EU rules make AI text easier to identify.
With GPT-6, ChatGPT can also show answers as charts or small tools
An answer can now be a small tool as well as text.
OpenAI has started rolling out GPT-6 in ChatGPT. Paying users got GPT-6 Sol on 7 October. A day later, users with a free account or a Go subscription followed. They got the smaller GPT-6 Luna. The new Intelligent UI feature can show answers with charts, buttons, forms and small tools. UI stands for user interface: everything you see and click on screen. Think of a calculator or a tool for splitting costs. For office work, this kind of answer can sometimes be more useful than a long block of text. In Enterprise, administrators decide whether the feature is on. If you do not want to see the visuals, you can turn them off on the web using the Layout and visuals setting. So the rollout changes more than just the model. It also changes what you see on screen. OpenAI also says the model starts answering while it is still thinking. For questions where it searches the web, OpenAI says this cuts waiting time by 44 per cent. That is compared with GPT-5.6 Instant. A chart or form can look like a finished result. But the format does not prove that the content is correct. You could try the feature on a simple task first. Then see whether the result is correct.
Background for subscribersChatGPT can now show information with charts, buttons and forms. This can make office work easier. But a polished display does not prove that the content is correct. This week there is also news about invisible watermarks for AI text in the EU. The other stories cover preparations for AI oversight and new tools in Gemini and Copilot. Two agreements can help: which AI services are allowed at work, and checking payment changes through a second channel.
AI text in ChatGPT gets an invisible watermark in the EU
OpenAI is adding an invisible watermark to text from ChatGPT and Codex. This applies to users in the EU, on all subscriptions. The rollout will take place over the coming weeks. The technology is called textGrain. It slightly changes the choice of words. This lets a detector recognise the pattern. A detector is a program that looks for a hidden pattern like this. OpenAI says the watermark remains when you copy and paste text. Editing is a weak point, though. If a quarter of the words change, OpenAI says recognition falls from almost 100 per cent to less than 20 per cent. For now, the detector is available only to approved researchers and organisations. So the absence of a watermark does not prove that a person wrote the text. At work, identifying AI text may become part of agreements about what you publish. The watermark is not a visible label. It works as long as most of the text stays the same. If you send messages on behalf of others, it may help to agree who checks the content.
Background for subscribersThis is a taster.
Subscribers read the whole Office management edition: every story, each with its background in plain language.
Subscribe →Not ready for that? Read the free letter first →
AI and tech, made simple.
This is the Office management edition. Subscribers get the background to every story, in plain language.
Subscribe
inklaretaal · Amsterdam · © 2026