Three stories from the past 48 hours that you will notice if you work with AI professionally. Grok text from the API will become recognisable as AI text from 7 November, an adviser to the European Court believes Gemini may infringe press rights, and the UWV sees AI as a cause of more highly educated people on unemployment benefits.
Grok API: an invisible watermark from 7 November
According to the xAI documentation, responses from grok-4.7, grok-4.6, grok-4.5, grok-4.3 and the three grok-4.20 variants will carry a watermark from 7 November. The reason is the European AI Act. The mark is built into the model's word choice and adds no characters, so copying and pasting does not remove it. Detection is statistical, not foolproof. You can see whether a response is marked in the `x-watermark` header. A team administrator can switch the watermark off per model in the SpaceXSI Console, but must then confirm that a legal exemption applies. Access to the detector is available on request only.
What does this mean for you? If you build a product on the Grok API, your output will soon be recognisable as AI text, just as with Claude and ChatGPT, which already received a watermark. If you switch it off, responsibility for the transparency obligation lies with you. You can read how Grok scores on privacy on the Grok page and in the GDPR check for Grok in software development.
Advocate General: Gemini may infringe press rights
In case C-250/25 brought by the Hungarian publisher Like Company against Google, Advocate General Maciej Szpunar concluded on 3 November that a chatbot such as Gemini may, in certain circumstances, infringe the rights of press publishers, for example if it uses articles behind a paywall or reproduces the original text in recognisable form. IEX reports this. According to him, training language models on protected material falls under the text and data mining exception. According to Dataconomy, this is the Court's first case on generative AI. The opinion is not binding; the Court will rule in the coming months.
What does this mean for you? If you use AI answers that rely on news articles in your own publications, check the original source and cite it. This applies especially to AI in media and marketing. What you can and cannot share in Gemini is set out in the GDPR check for Gemini in media and on the Gemini page.
UWV: more highly educated people on unemployment benefits, AI as a cause
The share of highly educated people among unemployment benefit recipients rose from 30 percent in the first half of 2022 to 38 percent in 2026, DutchNews reports. UWV director Judith Duveen increasingly sees AI as the cause. The professions named include programmers, marketing and advertising staff, market researchers, accountants, junior lawyers, translators and customer service staff. The financial services sector is expected to lose the most jobs.
What does this mean for you? If you work in one of those fields, it pays to learn how to work well with AI yourself. Start with the options for accountants, translation or customer service.
In brief
- Grok API text will carry an invisible watermark from 7 November under the AI Act.
- An EU adviser sees a possible infringement of press rights by Gemini; training falls under the TDM exception.
- The share of highly educated people on unemployment benefits rose to 38 percent; the UWV points to AI.
- All models and their privacy scores are at models and safety.