Urgent.News

What's breaking now, across thousands of outlets.

AI

Civil society groups push FTC to sue AI companies over book destruction

Deliberately destroying rare books might be the final straw in the eyes of the Federal Trade Commission.

Civil society groups push FTC to sue AI companies over book destruction

Earlier this month, civil society groups urged the Federal Trade Commission (FTC) to investigate AI companies for allegedly destroying physical copies of books to train their AI agents. The groups argue that this practice, reminiscent of book burning, could violate antitrust laws by raising costs for competing AI agents and denying them access to essential source material.

The groups include Demand Progress Education Fund, the Consumer Federation of America, and the Institute for Local Self-Reliance. They urge the FTC to view this practice as part of a broader pattern of anticompetitive behavior designed to create a barrier around AI incumbents. The destruction of older books, particularly those published before 2022, is especially concerning, as these books are unlikely to have been written by AI and are more likely to have been well edited.

The groups also note that this practice undermines the concept of knowledge as a utility, likening it to selling electricity or water, when it involves the wholesale destruction of books, a primary repository for human knowledge.

Written by urgent.news from Mashable's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at mashable.com →

More in AI

LLM Model Fingerprinting: Verify What Your AI Gateway Is Really Serving

Your prompt can ask a model what it is. Your production system should not trust the answer. A model can say it is GPT, Claude, Gemini, Llama, Qwen, or anything else.

  • LLM model fingerprinting verifies AI gateway serves expected model and runtime.
  • Traditional prompt-based identification unreliable due to model behavior changes.
  • Effective fingerprinting uses tokenizer behavior, API validation, runtime metadata.

More from Saturday 22 August →