Three posts from recent days. ChatGPT can now access your iCloud mail and calendar, four major AI companies will respect a signal that keeps your website out of AI training, and a parliamentary majority wants to ban generative AI for primary school pupils.
ChatGPT gets access to iCloud Mail and Calendar
Since 27 October you can connect iCloud Mail and iCloud Calendar plugins in ChatGPT, on the web, in the desktop app and on iOS, and on all plans, including the free plan. According to 9to5Mac, ChatGPT can then search mail, summarise your inbox and view and change appointments. Sign-in works with an app-specific password from Apple. It only works with addresses at @icloud.com, @me.com and @mac.com.
Notebookcheck points to a pitfall: if you disconnect the link in ChatGPT, that app-specific password remains valid with Apple. It does not expire on its own. You revoke it yourself via account.apple.com, under Sign-In and Security.
What does this mean for you? If you use an iCloud address for work, for example as a freelancer, it often contains personal data of clients. If you connect ChatGPT, OpenAI processes those emails. First check your plan and training settings on the ChatGPT page, and revoke the password with Apple if you stop. If you want to compare with an alternative, see ChatGPT vs Claude.
One signal in robots.txt for AI training
OpenAI announced that it will respect a shared signal in robots.txt, allowing you as a website owner to indicate whether your content may be used for generative AI training. The signal comes from the IETF's AI Preferences working group. According to Thurrott, Anthropic, Google and Microsoft are also taking part. Anthropic has already implemented it, and the other three will follow in 2027. It is separate from search indexing: you can remain findable in search engines and still say no to training. OpenAI's existing per-crawler rules will continue to work.
What does this mean for you? If you manage a website, for example as a publisher or agency, you will no longer need a separate rule for each AI provider. Until 2027, those individual rules will still be needed for most providers. More on AI in this field can be found at AI for media and AI for marketing.
Parliamentary majority wants generative AI out of primary schools
During the debate on the education budget, Laurens Dassen (Volt), together with the CDA, submitted a motion for a ban on generative AI for pupils in primary education. According to NOS, ten parties with a combined 88 of the 150 seats support the proposal. State Secretary Judith Tielen is not in favour of a ban for now and is first working on a plan for responsible use of AI in education, reports Computable. A motion is not binding: the cabinet decides itself how to carry out an adopted request.
What does this mean for you? If you work in primary education, the proposal concerns use by pupils, not learning about AI. Check which lesson platforms have pupils work with generative AI themselves and which use AI only for the teacher, such as the text generation in Gynzy. An overview is available at AI for education, and for general-purpose models in the classroom, the GDPR check for ChatGPT in education and the GDPR check for Gemini in education.
In brief
- If you connect iCloud Mail to ChatGPT, revoke the app-specific password with Apple yourself when you stop.
- A single robots.txt signal against AI training is coming: it already works at Anthropic, and will come to Google, Microsoft and OpenAI in 2027.
- A parliamentary majority supports a ban on generative AI for primary school pupils, while the state secretary wants a plan first.
- All models and their privacy scores are at models and safety.
