fb tracking

LLMs in Search Engines Harm Competition, Content Creators

WASHINGTON, D.C. – Federal lawmakers and antitrust regulators should launch an immediate investigation of large language model (LLM) incorporation into search engines and issue guidelines on anti-competitive practices, Public Citizen said in a letter sent today to the Federal Trade Commission, the U.S. Department of Justice, and the chairs of the House and Senate antitrust committees.

“LLM-provided search threatens to sabotage the open internet, shifting still more power to Google and Microsoft, at the direct expense of content providers,” said Robert Weissman, president of Public Citizen and author of the letter. “LLM-provided search results take and synthesize information available on the internet, decreasing the likelihood that users will click through to links of the original content generators and providers.”

Google, Microsoft, and other firms are incorporating LLMs into their internet search functionality. The prospect of Google integrating its new Gemini AI into its standard search function in early 2024 makes this request especially urgent. LLM incorporation into search engines may unfairly and substantially injure competition. Even more profoundly, it could enable dominant search firms to effectively enclose and privatize the open internet.

The letter highlights a number of examples that shows why users are unlikely to click through to other sources after being given detailed, narrative answers in a search reply, and how LLM-authored search replies sometimes borrow heavily from other sites.

“It will be increasingly difficult for content providers to monetize their investments – or for nonprofits to gain followers or for volunteers to get credit – if users get all the information they are seeking from search without clicking through to the content providers’ websites,” the letter notes. “[T]he incentives to develop and innovate web-based content will diminish still further, threatening the vigor and even viability of the open internet.”