How an Adult Performer’s Copyright Claim Led Google to De-Index a Climate Science Blog

Last week, Google delisted webpages of two environmentalist sites after allegations of copyright infringement. One casualty was a 2023 article about changes in Atlantic Ocean currents from RealClimate.org — which shares “climate science from climate scientists,” according to its homepage. The other delisted webpage, from the website of the environmental nonprofit Sierra Club, explained how plastic paint pollutes oceans.

The two articles had not copied anyone’s work. Yet the source of the claims against them did not seem to be a partisan activist or a covert campaign from the fossil fuel industry.

Instead, the takedowns arrived on behalf of an OnlyFans performer.

On July 19, someone acting for Ocean Larsen, the performer, sent a copyright takedown notice targeting a number of other adult sites, presumably ones which legitimately contained pirated content of Larsen’s. But mixed in with those sites were a number of links whose only connection to Larsen seemed to be that they had “ocean” in their URLs. This included the RealClimate.org blog post, the Sierra Club article, and a number of other news stories related to oceans (these other links were not delisted, according to a Google transparency report.)

It seems highly unlikely, even impossible, that a human could have made such a blatant mistake. The error instead strongly suggests the process that created the DMCA notice was entirely automated, with some sort of program using keywords to find potentially-infringing URLs. This phenomenon — of legitimate articles getting swept up in a list of allegedly infringing links about adult performers — is surprisingly common, like one from June on behalf of OnlyFans performer Summer Blayne, which resulted in Google de-indexing two L.A. Times articles.

Still, the exact process remains somewhat baffling. If a DMCA notice creation program was simply searching the entire internet for URLs containing the word “ocean,” why did it only turn up these specific articles and not the potentially thousands of others?

In any case, decades after the introduction of the DMCA, the law’s operation in the wild still results in these unintended yet baffling outcomes. Errors like these highlight the need for not only greater transparency in general regarding takedown notices of whatever sort, but also for more research to be done on both notices and on trends within the global takedown ecosystem. Lumen looks forward to continuing to contribute to both.