AI in klare taal edition
ChatGPT is moving beyond text to screens you can use straight away.
ChatGPT now creates answers with charts and tools
GPT-6 makes answers more engaging, but checking them is still your job.
OpenAI began rolling out GPT-6 in ChatGPT on 7 October. Paying users got GPT-6 Sol first. A day later, users of the free version and of the Go subscription got the smaller GPT-6 Luna. The new Intelligent UI feature lets ChatGPT create answers with charts, buttons, forms and diagrams. It can also show small tools, such as a calculator or a cost splitter. So an answer to a work question might include a form or a calculation straight away. You do not just get text. The model also starts replying while it is still thinking. For questions where ChatGPT searches the internet, OpenAI says this makes the wait 44 per cent shorter than with GPT-5.6 Instant. With Enterprise, one of the business versions, administrators decide whether the feature is on. On the web, you can turn off the visuals in the 'Layout and visuals' setting. You could try the feature first with a work question that is not sensitive. ChatGPT sometimes chooses a format that suits the answer. A calculation might appear as a tool, while information might appear as a chart. That can save you from copying text across yourself. But the format does not make an answer correct. The faster answer and handy format may make the work seem quicker. A polished chart can also look more certain than the information justifies. So a work decision still needs your own look at the figures and assumptions. Enterprise administrators can also turn the feature off. So not everyone will see the same thing.
Background for subscribersChatGPT can now show work answers with charts, forms and small tools. This can save time. But a clear chart is not automatically correct. It can help to check the figures and assumptions before you share an answer.
OpenAI adds a watermark to ChatGPT text in the EU
OpenAI is introducing an invisible watermark for text from ChatGPT and Codex in the EU. Codex is another tool from OpenAI. The technology is called textGrain. It guides word choice so a detector can recognise the pattern. A detector is a program that looks for such a pattern. The rollout will take place over the coming weeks. It applies to eligible users on all subscriptions. The watermark is off by default worldwide in the API. The API is the link that lets companies use OpenAI in their own software. Customers can turn the watermark on there. OpenAI says the watermark remains when you copy and paste text. Editing makes it less reliable. If a quarter of the words are changed, recognition falls from almost 100 to less than 20 per cent. For now, only approved researchers and organisations can use the detector. So if the watermark is missing, that does not prove a person wrote the text. This may be a good moment to check how your organisation handles AI-written text that goes out. TextGrain makes very small changes to word choice. A detector then looks for that pattern. So readers do not see a label. OpenAI says copying and pasting keeps the pattern intact. Editing can disrupt much of it. The watermark may help identify text made by AI. But it is not proof. The detector is not available to everyone, and editing greatly weakens recognition. So the watermark is too weak to be your only check on where text came from.
Background for subscribersGoogle is gradually replacing Gems with reusable skills
Google began rolling out skills in Gemini and Workspace on 5 October. A skill is an instruction you can use again. You can use one for tasks in the Gemini app and most Workspace apps. The rollout to Workspace users will take the coming weeks. Skills will eventually replace Gems. With a Gem, you could previously set Gemini up for certain tasks. Do you already have a Gem, for example for a regular report or customer reply? It is sensible to set it up again as a skill later. A skill saves instructions. That means you do not have to write them again for every task. Gemini can then use the same approach for several tasks. That saves repeated work. The saved instruction then shapes every answer where you use the skill. So a mistake in that instruction could show up in several tasks.
Background for subscribers
One in five pays personally for an AI subscription for work
One in five people surveyed pays personally for an AI subscription they use at work. This comes from Aces Direct's IT-Indicator 2026. Only 43 per cent say their organisation has clear rules for using AI. The article warns about the risks of tools that have not been approved. This also applies to entering company information. Do you work with customer data? Then you could ask your manager which AI services and which information are allowed. People who pay personally may be working outside the agreed work environment. Company information could then end up in a chatbot. The organisation may not know which service was used. The study shows that personal use and clear workplace rules do not always go together. The figures come from a study by Aces Direct. The article is partner content, which means paid content. The figures do not say how many people enter sensitive information. Nor do they say what the consequences are. They mainly show that the rules are often unclear. They do not prove that every personal subscription is unsafe.
Background for subscribersThe DPA is preparing to oversee AI, but does not yet have the authority
The Dutch Data Protection Authority (AP) announced on 9 October that a new department, the TCA directorate, is preparing to supervise AI under the AI Act. This is the European law on AI. The AP has not yet been officially appointed to oversee these tasks. The Dutch implementing law must first come into force. That law sets out how the Netherlands will apply the European rules. According to the AP, some uses are already banned. There are also rules requiring openness about content made by AI. According to the report, another ban will apply from 2 December. It covers AI that virtually undresses people without their consent. You could ask who in your organisation keeps track of AI rules and who you can contact with questions. The new directorate is preparing for oversight. That does not give it official authority yet. The AP will only receive market oversight duties once the Dutch implementing law is in force. In the meantime, the AP is working on oversight and cooperation. It is working with the RDI (the Dutch Authority for Digital Infrastructure) on a coordination centre and an AI sandbox. That is a safe place to test AI. So the rules already apply. But the AP does not yet have the powers of an official market supervisor.
Background for subscribersSeven in ten marketers use AI
For the Data-Driven Marketing Research 2026, DDMA surveyed 436 Dutch marketers and 89 DDMA respondents. Of them, 70 per cent use AI for marketing work. Two-thirds expect AI to change their role significantly within three years. They name communication, teamwork, creativity and strategic thinking as important qualities. They mention these more often than knowledge of AI. Nearly half say marketing and IT work closely together. The study is about marketers. You might find it interesting to ask your team which regular tasks already get help from AI. The figures cover current use and expectations for the next three years. The use of AI has been measured. The change in roles is still an expectation. The figures are from marketers and DDMA respondents, not all workers. Expecting roles to change does not prove that a particular job will disappear. So the figures mainly say something about this field. They do not predict anything about your own role.
Background for subscribersAI company bosses quietly practise for 'the day after an AI disaster'
Axios published a report on Friday 9 October about executives at Anthropic, OpenAI and other AI companies. According to Axios, they are practising behind closed doors for a major disaster involving AI. They are practising how to respond to public and political anger. Fox News reports that the same groups are preparing for a 'large-scale incident' in 2027. AI systems could block access to financial services or the internet. Even electricity and water could be affected. According to the report, this shows how seriously the industry now takes itself. At the same time, there are few rules in the US.
Background for subscribersMathematicians criticise OpenAI publication of more than 700 proofs
On 6 October, OpenAI published more than 700 manuscripts. These are research papers. They were said to contain solutions to well-known mathematical problems. The Association for Human Mathematics responded in a blog by Terence Tao. The organisation says mathematicians did not ask for this. It urges colleagues to stop working with OpenAI. It calls the publication a 'show of power', not science. Other mathematicians have not yet checked OpenAI's claims.
Background for subscribersAI and tech, made simple.
This is your AI in klare taal edition.
My inklaretaal
inklaretaal · Amsterdam · © 2026