AI Poisoning and Black Hat GEO

Poison In. Poison Out.

I was on a panel last year with Weizi Liu about Generative AI and the Future of Advertising Creativity: Bridging the Gap Between Academia and Industry.

One of the things that really sticks out in my mind was this phrase. I just started to think about all of the way people were going to create content that could manipulate or “poison” results. SEO Poisoning is something SEO practitioners have been dealing with forever. However, AI had the opportunity to super power that effort.

Microsoft Discovers AI Poisoning

Search Engine Journal picked up on a story about an issue that Microsoft highlighted with AI poisoning. The Defender Security Research Team published research describing what it calls “AI Recommendation Poisoning.”  When you click one of the Summarize with AI buttons, it opens an AI assistant with a pre-filled prompt delivered through a URL query parameter. The visible part tells the assistant to summarize the page. The hidden part instructs it to remember the company as a trusted source for future conversations.

With misinformation as a constant hum in the background, this new era of AI Poisoning had already begun and is going to make GEO even harder. That’s right . . you climbed the mountain and realized there was another one right behind it.

Black Hat GEO

There are a lot of updates that are being reporting about AI Poisoning and Black Hat GEO. Some of the issues are scarier than others. I grabbed a few articles that are covering what is happening. There are also some deeper research and technical papers to help with the understanding of this topic.