1 November 2026
NLA Media Access's 'Collective Intelligence Advisory Service'. The Bot That Hid on One Site Can't Hide on All of Them.
FIELD GUIDE: THE INVISIBLE SCRAPER
How to find it. Where it lives. What it does when you look.
Distinguishing features
A stealth bot looks like a reader. It arrives with a residential IP address, loads pages at roughly human speed, and ignores the robots.txt file on the doorstep. About 30% of AI scrapes in Q4 2025 ignored explicit robots.txt instructions1. The bot knows the rule. It just doesn't follow it.
By end of 2028, a UK court will use cross-site traffic logs as the main evidence in a copyright case against an AI scraper. If no such case files, the NLA's 18-month clock ran out quietly.
Habitat
Hardest to see on any single site. A scraper that looks like a heavy reader on one title gives itself away when you check it across 14,0002. Visit a thousand pages at the Telegraph, pause half a second between each, then show up in identical form at a regional news site an hour later. One site sees a strange afternoon. Fourteen thousand sites see a machine.
Behaviour when caught
Some ask for a licence. The NLA already sells scraping licences with tight limits on use2. The ones found in the pilot with Newsquest and The Telegraph, running from Eastern Europe and the Asia-Pacific, were not waiting to talk terms.
Identifying marks
The cross-site pattern. NLA Media Access's Collective Intelligence Advisory Service, launched 30 October 2026, pools web server and CDN log data from participating news titles3 to find what no single title can see alone. Its existing Text Tracker removed more than 150,000 infringing items this year, including over 40 whole sites taken down4. The new service goes after what that missed: the visitor who looks fine at your door but has been into every house on the street.
What the field guide doesn't tell you
A bot that buys enough rotating residential proxy addresses can still stay thin enough to evade this. A Piano analysis of Cloudflare data found 70% of bot traffic is unverified5. Nearly 40 scraping vendors sell exactly this kind of cover1. The NLA catches the scraper running at full tilt. Not the one that paid to go quiet.
The thing everyone knows and nobody says
robots.txt was a note on the door, not a lock6. News titles knew this in 2023. The NLA's new service is the admission that the note failed.
The strange part: news titles are now doing to scrapers what AI companies did to the web. Pooling data across many sources to find a pattern no single source could see. They borrowed the method from the thing they're trying to catch.
Whether a court finds that useful is the 18-month question.