Last week an AI Review’s Google search results used one of my WIRED articles in an unexpected way that makes me fear for the future of journalism.
I’ve been experimenting with AI Reviews, the company’s fresh generative AI feature designed to answer online queries. I had asked him a lot of questions about topics I had been discussing recently, so I wasn’t shocked to see a link to my article as a footnote at the bottom of the box that answered my question. But I was surprised by how much the first paragraph of the AI Review was taken directly from my text.
The screenshot below on the left is from an interview I did with one of Anthropic’s product developers about tips for using Claude’s chatbot. The screenshot on the right is part of a Google AI overview that answers a question about using the Anthropic chatbot. Reading both paragraphs side by side, I am reminded of the cheater in class who copied the answer from my homework and barely bothered to change the wording.
Reece Rogers via Google
Without AI reviews enabled, my article was often a featured snippet at the top of Google search results, offering a prominent link that curious users could click on when looking for advice on using the Claude chatbot. During my initial tests of Google’s fresh search engine, the article snippet still appeared for relevant queries, but it was moved under the AI review answer, which was pulled from my reports and placed its aspects into a 10-item bulleted list.
In email exchanges and phone calls, a Google spokesman acknowledged that AI-generated summaries may apply text snippets directly from websites, but defended the AI reviews as clearly referencing original sources. Well, in my case, the first paragraph of the answer is not directly attributed to me. Instead, my original article was one of six footnotes hyperlinked at the bottom of the results. Given that the source links are so low down, it’s strenuous to imagine any publisher generating significant traffic in this situation.
“AI Reviews will conceptually match information appearing in the most popular search results on the Internet, including those linked to in the review,” a Google spokesperson wrote in a statement to WIRED. “This information does not replace web content, but is intended to help people understand what is out there and click to learn more.” Looking at the word choice and overall structure of this AI review, I disagree with Google’s characterization that the result can only be a “conceptual match” with my text. It goes further. Moreover, even if Google developers did not intend for the feature to replace the original work, AI reviews provide direct answers to questions in a way that conceals attribution and reduces the incentive for users to click through to the source material.
“We see that links included in AI reviews receive more clicks than if the page appeared as a traditional web entry for that query,” a Google spokesperson said. WIRED hasn’t provided any data to support this claim, so it’s not possible to independently verify the impact of AI features on click-through rates. It’s also worth noting that the company compared AI referral traffic to more classic blue link traffic from Google, rather than articles selected for snippet with a response where the rates are likely high. higher.
While many AI lawsuits remain unresolved, one legal expert I spoke with who specializes in copyright law was skeptical that I could win any hypothetical case. “I don’t think you would have a strong case for copyright infringement,” says Janet Fries, a lawyer at Faegre Drinker Biddle & Reath. “Generally, copyright law tries to stay out of the way of things that are useful and helpful.” Her perspective focused on the type of content in this particular example of original work and explained that it is quite hard to argue that instructional or fact-based writing, like my advice column, is compared to more artistic work, like poetry.
I’m definitely not the first person to suggest focusing on your target audience when writing chatbot prompts, so I agree that the factual aspect of my writing complicates the overall situation. But it’s strenuous for me to imagine a world in which Google gets to this exact paragraph about Claude’s chatbot in its AI review results without first referencing my work.