AI update for (Christian) educators, April 2026

At the end of each month (or the start of the next 😅) I look back: what happened that educators, in my view, should know? A brief selection of AI news that touches on family, school and development. Now then, April 2026.

1. Lots of new models

April was perhaps the busiest month ever for new AI models. GPT-5.5, Claude Opus 4.7, Grok 4.3, DeepSeek V4. For parents, it is mainly relevant that all these free models keep getting more powerful too. The threshold for children to reach very powerful AI gets lower every month.

2. Parents get more insight into teens' AI conversations (Meta, April 2026)

This month Meta rolled out a new feature on Instagram, Facebook and Messenger: parents who have enabled supervision can see, per week, which topics their teen has talked about with the AI chatbot. Think of categories like school, health or lifestyle. Work is also underway on alerts for conversations about, for example, self-harm.

Although this is a useful step, it deserves a caveat: many parents probably don't even know their teen is already having chatbot conversations through, for example, WhatsApp or Instagram. You can check whether the supervision feature is enabled; young people turn out to bypass these features fairly often.

3. OpenAI publishes child safety plan (8 April 2026)

OpenAI released an extensive policy framework specifically aimed at protecting children. The reason is the sharp rise in AI-generated abuse material involving minors: in the first half of 2025, more than 8,000 cases were reported worldwide, 14% more than the year before. OpenAI is calling for legislation that makes this material more explicitly punishable and for better cooperation with law enforcement.