Three posts from this week that bear directly on how you use AI at work. The Dutch Data Protection Authority (AP) has set up its own directorate for AI oversight. Anthropic is updating the usage rules for Claude as of 12 November. And since 8 October, Claude can build live dashboards from your business data.
AP prepares AI market supervision
The Dutch Data Protection Authority (AP) has started the directorate for AI/Algorithms Supervision and Coordination (TCA), the AP reports. The new directorate succeeds the Algorithms Coordination directorate and continues the coordinating supervision the AP has carried out since 2023. In addition, together with the Netherlands Authority for Digital Infrastructure (RDI), it is preparing market supervision under the AI Regulation, as stated in the directorate's description.
The AP is the intended market surveillance authority, but it will only receive those powers once the Dutch implementing act for the AI Regulation takes effect. The rules themselves already apply, such as the ban on certain applications and the transparency obligation for AI-generated content.
What does this mean for you? Expect a single clear point of contact that actively monitors compliance. If you work in the public sector, in healthcare or at a bank, record now which AI systems you use and for what purpose. At /toezicht/ you can see which regulator covers what.
New usage rules for Claude from 12 November
On 8 October, Anthropic published its annual update to the Usage Policy. The rules apply to all users, including if you use Claude via the API or a business subscription. The key points:
- High-risk use: the requirements have not changed, but they have been clarified. For decisions concerning health, legal rights, finances and access to essential services, a qualified human must be able to review the outcome. The person concerned must know that AI has been used.
- Surveillance and investigation: tracking people without their consent is not allowed, including after the fact through collected data. Claude may not decide or advise on who is investigated, arrested or prosecuted. Fraud monitoring, content moderation, journalism and legal research remain permitted.
- Misleading campaigns: a new section brings together the rules against fake accounts, hidden sponsors and influence campaigns.
- Conduct towards the model: persistent and needlessly cruel behaviour directed at Claude is prohibited. Ordinary frustration or testing does not fall under this.
If you use Claude as a lawyer, a financial adviser or in healthcare, make sure a human checks every assessment and that your client knows AI was involved. That also fits the GDPR rules on automated decision-making.
Claude builds dashboards from your business data
On 8 October, Anthropic launched Claude Dashboards, in beta for paid subscriptions. You connect a data platform such as BigQuery, Snowflake, Databricks, Amazon Redshift or ClickHouse, or a connector such as Salesforce, and ask in plain language for a dashboard that updates as the data changes. Claude Motion makes animated explainers as video and is in beta for Team and Enterprise. Docs, Slides and Design are out of beta and available on every plan, including free. The standalone Claude Design app follows on 14 December.
Before you connect business data, check the terms that apply to your subscription. The Claude page and /veiligheid/ set out, per subscription, whether your input is used for training.
In brief
- The Dutch DPA has a directorate for AI supervision; genuine enforcement powers follow with the implementing act.
- From 12 November, stricter and clearer usage rules apply to Claude, including for business and API users.
- Claude Dashboards is in beta for paid subscriptions. Check how your data is processed before you connect business data.
- Compare models at /modellen/.