[Digital Today reporter Chi-gyu Hwang (황치규)] Tech platform companies are launching a broad cleanup amid worries that AI-generated content is flooding the internet and polluting the information ecosystem.

A report by The New York Times on Aug. 17 (local time) said services ranging from mobile messengers and restaurant review sites to academic archives and dating platforms are stepping up efforts to filter out or cut exposure to low-quality AI-made content.

Streaming music service Spotify deleted 75 million items, including mass uploads, duplicate tracks and other spam content. That is a significant share of the platform's overall portfolio.

LinkedIn also developed a button that allows users to report low-quality content easily made with AI, so-called AI slop, calling it a "top priority for all of us" in July.

Pinterest updated its system in October last year so users can set limits on how much AI-generated posts they see. TikTok is testing in the United States a feature that lets users increase or reduce the amount of AI content shown on screen. TikTok also automatically labelled more than 3 billion AI-generated videos and deleted more than 377,000 videos in the first quarter for violating AI-related policies.

Google researchers at YouTube's parent company removed 130,000 channels from the platform after they posted low-quality content over six months.

Slop has become a word symbolising an era in which AI is spreading rapidly. Dictionary publisher Merriam-Webster also selected slop as its 2025 word of the year.

French music streaming service Deezer said this spring that almost half of the tracks registered on its system each day were generated by AI, and that a significant number of them had never been listened to even once by users. The New York Times reported that more than one-third of content uploaded to Apple Music is also generated by AI.

Aiden Walker (에이든 워커), a content creator and researcher who studies online trends at the Carnegie Endowment for International Peace, a think tank that studies international issues, said, "People are increasingly tired of the effects AI has on the environment and mental health and the pressure it puts on jobs," adding that "AI slop is also another festering wound."

He explained that as user complaints about AI slop grow, public opinion has visibly shifted and companies have finally begun to respond.

Technology companies are focusing on reducing or eliminating low-quality clickbait content that is easy to make and spread but quickly forgotten as part of platform cleanups.

Major tech companies have actively supported users in producing content with AI.

An experiment conducted last year by AI-based online video editing service Kapwing found that more than 20 percent of the short videos YouTube first shows to new users were low-quality, AI-generated content. Kapwing estimated that one of YouTube's most popular AI slop channels earned $4.25 million a year, and The New York Times reported the channel was later suspended from YouTube's monetisation programme. Kapwing data from June this year showed that 6 of every 10 videos posted to short-form video platform TikTok were AI slop.

Six out of 10 TikTok videos are AI trash... YouTube also at 21 percent

Meta in July released an image-generation tool that automatically enrolled public Instagram account content so image-generation AI could use it as a reference, but scrapped it three days later after a strong backlash in what became an uproar.

Paul Shovlin (폴 쇼블린), an expert who researches AI at Ohio University, said, "AI slop is a double-edged sword. Many platforms saw a boost in user engagement thanks to AI slop," adding, "But at some point it started to become a burden. As too much low-quality content piled up, it became difficult for users to find high-quality content with the structure they want."

It is not easy for tech companies to examine AI slop content one by one and decide whether to leave it on a platform or take it down. They are therefore actively using AI technology to identify AI slop content as well.

Google researchers, rather than evaluating YouTube videos one by one, analysed patterns such as upload speed and repeated content formats to find signals of organised abuse. They found and deleted a cluster of 50,000 accounts.

Some digital media experts worry that this approach could be applied too broadly and disadvantage even legitimate creators who produce a lot of content, The New York Times reported.