Hi there.

I am writing this as I travel back to Europe to visit friends and family this summer. I’ve found myself reflecting on the first part of the year and realise there are a couple of conversations I found myself having over and over with news publishers.

So this newsletter dives into one thing I am convinced every publisher needs to do today when it comes to AI, as well as a glimpse into the new ChatGPT story tiles. 

If you are in the northern hemisphere like me, I hope you are having a great summer. 

Thanks for reading, 

Jodie

The one thing you need to do now on AI

There’s so much to think about regarding AI right now: tools, ethics, regulation, partnerships. But if I could make one recommendation for immediate action, it’s this: Measure and manage the bots that are scraping your sites. Please. 

Why? Because if you let AI companies take your content for free, you’re undermining any future business model for licensing or compensation.

As I wrote in an INMA report on working with AI companies earlier this year: Robots.txt does not cut it; it’s a gentleman's agreement at best. Many companies, such as Perplexity, will tell you they don’t scrape. Which is true. But it is also true that they buy data from third parties that do. 

And yes there are a LOT of third parties. Check out this list of AI bots that may be scraping your site. 

You may have seen Axios’ recent piece that laid out the stark reality of click-throughs as you can see on the chart below, showing the number of Web pages crawled per visitor referral.

These numbers should stop any news organisations in their tracks. The scraper bots are coming, and they’re taking a lot more than they give back.

In a recent INMA webinar, Robert Hahn from The Guardian shared how they are approaching AI licensing — and made it clear how surprised he was that so few publishers are acting on this. On that call, 59% of attendees said they are not blocking scraper bots.

Let’s be blunt: If you allow scrapers to crawl your content, you’re giving it away for free. And if it’s free, why would anyone pay for it?

But here is what you can do.

Companies such as Cloudflare, Tollbit, ScalePost, Dark Visitors, and Miso are all building solutions for this:

• Cloudflare is working on a tool to manage this at a large scale with blocking by default.
• Tollbit, ScalePost, and Dark Visitors are helping news publishers track and monetise AI usage for free.
• And Miso even has a tool that lets you see, right now, how many scrapers are crawling your site. (Warning: The results may surprise you.)

If you’re looking for more context or next steps, you can check out this earlier post. But the headline remains: This is the single most important first move for protecting your content in an AI world.

And if this resonated, please consider joining us at Media, Tech & AI week, where we’ll spend some time really diving into this.

Date for the diary: August 12 Webinar on mobile video and personalised content

What’s driving award-winning mobile formats and fast-moving product teams? 

In this INMA members-only Webinar, go behind the scenes with VG and HT Labs to learn how they reimagined mobile video and personalised content discovery — and came away with INMA’s top product honours at the recent Global Media Awards.

This session will share the tools, tactics, and team structures that set these innovations apart. Sign up here.  

ChatGPT’s new publisher-friendly tiles 

Have you noticed ChatGPTs new tiles? These look very publisher friendly in that they encourage click-through to sites.

I had heard about them, but it took me a few attempts to get them to show up. In fact, I’ve only been able to get them to show once. The query that worked: I asked for “latest updates on fashion shows,” assuming that latest and shows would trigger specific, recent references.  

Then I wondered how they were triggered and thought there is no better way than to check the source. So I asked ChatGPT. This may not be exactly right for these types of tiles, but this looks directionally right. (For reference, I also reached out to OpenAI to ask about this.)

• Contextual triggers: Whenever you ask a question or make a request, the system analyses your last few messages for key intents (e.g. asking for options, next steps, clarifications). If it detects that you might benefit from a quick follow-up — like choosing between a few ideas, confirming details, or requesting something related — it will surface one or more suggestion tiles.

• Intent and entities: The model looks for actionable items (verbs like “choose,” “compare,” “show me”) or entities (products, dates, places) that lend themselves to further exploration. For example, after you asked about chair options, tiles might suggest “Show me more chair styles” or “Compare materials” because those are logical next moves.

• Dynamic generation: These tiles aren’t hard-coded. They’re generated on the fly by the same underlying language model that’s replying to you. It predicts which short prompts would most likely help you continue the conversation smoothly.

• User interaction patterns: Over time, the system learns which types of suggestions users tend to tap. That feedback helps fine-tune which tiles get offered in which situations — so you’ll see more relevant prompts the more you interact.

What can you do to maximise your chance of tiles? I suspect that comes down to whether or not you optimise for AIEO/GEO (have we agreed an acronym on this yet?)

I have no doubt this improved click-throughs, but we don’t know how much. 

The real challenge here is that it makes the question of blocking bots and/or optimising for searching a little more grey. Should you make all or a subset of your content accessible? At the moment, these tiles are only appearing on a fraction of answers. 

We’ll be keeping an eye on this. If you have any showcases, lessons, or stats from the ChatGPT tiles, please share them with me (confidentially) at jodie.hopperton@INMA.org.

About this newsletter 

Today’s newsletter is written by Jodie Hopperton, based in Los Angeles and lead for the INMA Product and Tech Initiative. Jodie will share research, case studies, and thought leadership on the topic of global news media product.

This newsletter is a public face of the Product and Tech Initiative by INMA, outlined here. E-mail Jodie at jodie.hopperton@inma.org with thoughts, suggestions, and questions. Sign up to our Slack channel.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .