I update my AI in Marketing toolkit every few months because the AI landscape doesn't sit still. This is the fifth revision, and two sections needed real work: copyright and Answer Engine Optimization. If you've downloaded or received an older version, get the most recent AI in Marketing Toolkit update. The two sections to re-read are copyright (pages 6-9) and AEO (pages 17-19). New to the toolkit? Start at the beginning!

Here's what's different and what you should actually do about it.

Copyright

The legal picture has gotten clearer since the last version of my toolkit. Not settled, but clearer.

A few things to know:

• The U.S. Copyright Office said in January 2025 that AI outputs can only be copyrighted when a human contributes real creative work. Typing in a prompt and publishing what comes out doesn't count. The more your team shapes the final piece, the stronger your claim to it.
• How you label AI-assisted content matters. If you slap "AI-generated" on something your team actually wrote and edited, you could weaken your own copyright claim. A better label: "developed with AI assistance and substantially edited by [name or team]."
• The NYT v. OpenAI case is still going. Discovery wrapped in late 2025. The big question (is training an LLM on copyrighted content fair use, and more important, is your AI chat log discoverable?) is heading toward summary judgment. Worth watching, but not worth waiting on.
• The Getty v. Stability AI ruling came down in November 2025. It's significant because it's the " ...first UK ruling to address how copyright and trademark laws apply to the training and deployment of generative AI models" [source] The UK court largely sided with Stability AI, so on the image side, the law isn't moving the way rights holders hoped. Getty is appealing.

So what does any of this mean for your afternoon?

Two things, really. First, check the terms of service on the AI tools your team uses, and turn off any setting that lets your inputs train the model. Most paid tools have this setting; you have to flip it yourself. Second, be intentional about what your team uploads. Member data, unpublished research, proprietary content; keep those out, or anonymize them.

That's it. The lawsuits will resolve when they resolve. What you can control is what goes in and how you label what comes out.

Answer Engine Optimization

If you haven't started thinking about AEO yet, now's the time.

People are asking questions of ChatGPT, Google's AI Overviews, and voice assistants instead of typing keywords into a search box. The content that shows up in those answers looks different from the content that ranks well in traditional search.

The new AEO section walks through it, but the short version:

• Write around questions, not keywords. "How do I…" "What is…" "Why does…"
• Lead with the answer. The first two or three sentences of a section do most of the work.
• Use semantic HTML and schema markup so AI systems can actually parse what you've published.
• Keep it current. Authorship, dates, citations all matter more than they did before.

Here's why this is good news for associations: you already produce the kind of content AEO rewards. You have SMEs, you have credible research, you have a body of work with dates and authors on it. You just need to structure and label it so AI systems can find it.

What didn't change

Most of the toolkit didn't need a rewrite. The guidance on what never to put into a generative AI tool is the same. That's worth saying out loud. The fundamentals of using AI responsibly aren't changing as fast as the headlines suggest. Guardrails, policies, training, and fact-checking: that's still the work.

I'll keep updating this as things change. If you want a heads-up when the next revision drops, subscribe to my newsletter at cathylada.com/insights.