Meta’s revised business terms let it crawl publisher sites by default from October 30, and publishers have only days to decide, object and start logging what is taken.
The News/Media Alliance, a trade group that says it represents more than 2,000 publishers, sent Meta an open letter on October 22, asking it to withdraw the crawling provision in its revised Commercial Terms before they take effect on October 30 and answer by October 26.
The clause lets Meta use scrapers and bots to crawl domains tied to a business’s products or services. Meta removed the word “train” after objections, but the text still cites developing and improving “advertising and business products,” which the alliance calls undefined. Businesses must also warrant their rights to the content and indemnify Meta against third-party claims, and continued use counts as acceptance.
The alliance wants talks on which products would use scraped content, a scraping opt-in separate from Meta’s business tools, which bots are involved, and how robots.txt, registration walls and paywalls will be respected. Its members say they do not consent.
The letter shows a routine terms update carrying a rights grant, and it shows the notice period is the only leverage publishers have.
I learned this running local news platforms inside a large cable and broadcast company: content holds its value only when the operator owns the platform and the audience data around it. A publisher that decides before October 30 what Meta’s business tools are worth against the rights now requested will negotiate from a clearer position than one that finds out afterward, and one that can show what was collected will be heard differently from one that can only object.
Two German publisher associations, which say they represent more than 9,000 media brands across Europe, asked the European Commission on October 20 to prohibit Meta’s new terms from taking effect until a full review is done. Their joint statement says Meta intends to automatically capture, copy and store the complete websites and archives of publishers, calls it “blatant theft,” and argues the terms may breach the Digital Markets Act clause on fair, reasonable and non-discriminatory access conditions. U.K. politicians separately pressed the competition regulator to pause the deadline.
The Operator’s Take: Regulators rarely move faster than an October 30 date, so plan for the terms taking effect while the complaints are still pending.
Legal · Industry Positioning Put a dated objection on the record with the relevant regulator now, citing the fair-access clause, so it predates the terms taking effect.
Around 300 authors filed individual copyright claims against Meta on October 19 in the U.S. District Court for the Northern District of California, and their lawyers plan further suits for close to 1,000 more authors in the coming months. The complaints allege Meta downloaded hundreds of thousands of e-books from shadow libraries and used them to train its AI models. Bloomberg Law reports statutory damages could reach $150,000 per infringed work; none have been awarded. In a separate proposed class action, a judge ruled against the authors because they had not provided enough evidence of market harm.
The Operator’s Take: Individual claims route around the class ruling and put the burden on each claimant to show market harm, which makes evidence the scarce asset.
Build · Detection & Telemetry Capture dated evidence of licensing demand and substitution for your own catalog now, because the earlier class ruling turned on the absence of market-harm proof.
A study by an AI search visibility firm tested ChatGPT, Perplexity, Google Gemini, Google AI Mode and AI Overviews with 30,000 prompts covering 20 questions about 100 U.K. towns and cities, producing 244,725 citations from 10,000 websites. According to the report, ChatGPT drew on council or government sources for 81.3% of questions and local news sources for 0.9%, and only one local title appeared among its top 20 sources for local queries. On crime questions, police force websites supplied 57.8% of answers. The study’s author said AI is not yet making up for the search traffic regional publishers are losing.
The Operator’s Take: Answer engines prefer the source of record, so local publishers need to be licensed as structured, verifiable data rather than hoping for citations.
Commercial · Economic Model Package structured local records, such as planning notices, court lists and meeting coverage, as a priced data feed instead of waiting on citations.