Creators demand compensation
Publishers, creators and parts of the media industry are pressing AI platforms to pay for training data instead of relying on scraped content, and that debate is moving into mainstream coverage. ( ). At the same time, publishers report rising difficulties separating genuine submissions from AI‑generated material, with examples of canceled books and increasing verification overhead. (cbc.ca)
Publishers, authors and artists are pressing artificial intelligence companies to pay for training data as disputes over scraped content move from court filings into mainstream media coverage. (timesnownews.com) Times Now said it held a discussion on April 11, 2026, as part of a campaign on how artificial intelligence apps use creators’ and news publishers’ work, and it placed the fight in the same stream as lawsuits already filed in the United States and Europe. (timesnownews.com) At the same time, publishers are dealing with the other side of the problem: manuscripts that may have been generated by software. Canadian publishers told CBC that screening submissions now takes more time, and the Mia Ballard novel *Shy Girl* became a high-profile example after Hachette canceled its United States release over suspected artificial intelligence use. (cbc.ca) The basic dispute is about training data, the giant libraries of text, images and audio that models ingest to learn patterns. Creators and publishers want licenses and payment for that material instead of mass scraping from the open web. (copyright.gov) The United States Copyright Office said in its May 2025 training report that using copyrighted works for generative artificial intelligence training can implicate copyright law, and it said emerging licensing markets weigh against a broad fair-use claim where licenses are feasible. (copyright.gov) That licensing argument is gaining force as artificial intelligence companies broaden their public-policy push. OpenAI published a 13-page paper on April 6, 2026, calling for new economic policies around advanced artificial intelligence, while Anthropic has expanded its Economic Futures program and launched the Anthropic Institute to study artificial intelligence’s social and economic effects. (openai.com, anthropic.com, institute.anthropic.com) The Guardian reported on April 12 that major artificial intelligence firms are funding policy papers and thinktanks as public skepticism grows, a sign that the fight over training data is no longer staying inside legal briefs and industry conferences. (theguardian.com) Publishers say the market impact is immediate. CBC reported that editors and agents are spending more time verifying whether submissions are human-written, adding cost and delay even before courts settle whether training on copyrighted work requires permission. (cbc.ca) Artificial intelligence companies have argued in court and in policy debates that training is transformative and can qualify as fair use, while rights holders say the systems copy expressive work at scale and compete with the originals without consent or payment. The Copyright Office did not endorse a categorical answer, but it rejected a one-size-fits-all fair-use defense. (copyright.gov) For creators and publishers, the fight is narrowing to a practical demand: if artificial intelligence systems are built on books, articles, art and archives, they want a licensing market that pays for those inputs before more scraped material and more disputed submissions flood the system. (timesnownews.com, cbc.ca, copyright.gov)
