Two new things on the site, then the practical follow-through to last issue's story.
There is now a bot on the Playbook. Ask it anything about the four domains, or the part most people find hardest: how to choose between two answers that both look right.
It answers in your language. Write to it in Hindi, Spanish or Arabic and it replies in the same language, keeping the AIGP terms in English because that is how the exam presents them. You can also speak your question and have the answer read back, which is useful if you revise on the move.
It is grounded in the Playbook's own material, the 100 essential facts and the 19 Master Rules, and it is told to say when something is outside what it covers rather than invent an Article number. Free, no email, no signup. Look for the button in the bottom right corner.
The podcast has a second series too. The first four episodes were revision, one per domain. The new one is about technique: the Master Rules, the traps the exam sets again and again, and how to decide when you are down to two options with the clock running.
With Article 50 now applying, covered last issue, here is the practical follow-through: the Commission has published the icons themselves, free to download in SVG and PNG.
There are three, and the distinction between them is the useful part. A basic icon, for when AI was involved in creating the covered content, or when you are using your own text label or an interactive second layer. A fully AI-generated icon, for content produced with no human-created elements and no editorial control beyond the prompt. And a partially AI-modified icon, for human-made content that AI has since altered. Each comes in four variations, black, white, and both at fifty percent transparency, so it stays visible on any background.
Now the sentence people are getting backwards. The icons are optional. The disclosure obligation is not. You may use your own label or an interactive second layer instead, but Article 50(4) still applies either way, and the Commission is explicit that using the EU icon does not by itself establish compliance.
The placement rules are where this gets operationally hard. The icon has to be clearly perceivable at the point someone is first exposed to the content, which is Article 50(5), with nothing overlaying it, embedded into the content rather than sitting next to it, and it must survive being reshared or downloaded. That last requirement is the one that will break most implementations. A badge on your web page is not a label on your image.
For those preparing for the exam, this is Master Rule 07 in the wild: transparency obligations are independent of risk classification. None of this makes a system high risk. Article 50 applies because of what the content is, not what tier the system sits in, and the exam likes testing exactly that distinction.
Missed an earlier issue? Every one of these is now on the site, in full and free to read: https://aigpplaybook.com/newsletter/
If you know someone preparing for the AIGP or building in this space, forward this to them. They can subscribe at aigpplaybook.com.
One thing before you go. If you have had to actually label AI-generated content at work, what broke first, the tooling, the workflow, or getting anyone to agree what counts as partially modified? Reply and tell me. I read every one, and the answers shape what I write next.
What is actually moving in AI governance, plus what it means if you are sitting the AIGP. No fixed schedule, only when something matters.