Transfermarkt now blocks AI by licence: as of 17 November 2026 its robots.txt points to an RSL licence file that prohibits any AI use of its data, and around 100 bots are on the block list. It's a price tag on the numbers fans ask ChatGPT about every day, with gaps.
Key facts
- As of 17 Nov 2026, Transfermarkt's robots.txt points to an RSL licence (license.xml) that prohibits all AI use via ai-all.
- Around 100 bots are blocked, including GPTBot, ChatGPT-User, ClaudeBot, PerplexityBot and agents like Operator and Nova Act; OAI-SearchBot and Google-Extended are absent.
- The same file applies on transfermarkt.de, .com, .at, .ch and .co.uk; the 28 April 2026 version had no License line.
Researched and written with AI, to Luigi's brief, positions and editorial rules. Automatically checked against the sources. How I work →
Ask ChatGPT what a footballer is worth and the answer often sounds a lot like Transfermarkt. Transfermarkt no longer wants to give that answer away for free. The platform's robots.txt now opens with a new first line: a pointer to a licence file that prohibits any AI use. Our AI Agent Radar picked up the new version on 17 November at 03:32 UTC.
What Transfermarkt's new robots.txt says
The facts, as the files read this morning:
- License line: At the very top,
License: https://www.transfermarkt.de/license.xmlpoints to an RSL file. It setsprohibits type="usage"toai-allfor the whole site. - What ai-all means: Per the RSL 1.0 spec, any use by AI systems: training, grounding, indexing and retrieval. A License line outside any user-agent group applies to every crawler.
- Block list: Around 100 bots sit in one group with
Disallow: /, including GPTBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, PerplexityBot, Perplexity-User, MistralAI-User, meta-externalagent, Applebot-Extended, Amazonbot, DuckAssistBot and agents such as OpenAI's Operator, Amazon's Nova Act and Devin. - Whole network: transfermarkt.com, .at, .ch and .co.uk serve the same file with the same licence pointer.
For comparison, the Internet Archive copy from 28 April 2026 already blocked GPTBot, ChatGPT-User and Claude, but had no License line and no agents like Operator or Nova Act. So the no to training isn't new. What's new is Transfermarkt declaring its data a licensed product.
The licence targets the AI answer, not just training
Looks defensive. It isn't. To me, it's a price tag. Transfermarkt sits on numbers nobody else has at this depth: market values, contracts, transfers. Exactly what fans type into a chatbot. And its owner knows what the deal looks like: Axel Springer, majority owner of Transfermarkt since 2008, agreed with OpenAI in December 2023 that ChatGPT would show summaries from Bild and Welt with a source link, in exchange for licence fees. Yesterday I wrote about the same mechanism with OpenAI and LVMH's media titles. Close the door first, then sign the contract.
That's the value transfer: whoever owns unique data stops selling clicks and starts selling citations inside the AI answer. Yahoo Scout already pays by contribution to the answer.
The catch: the door is only half shut
Three things make the signal softer than it looks:
- OAI-SearchBot, the crawler behind ChatGPT search, isn't on the list. Neither are Google-Extended or Claude-User.
- robots.txt and RSL are requests, not barriers. OpenAI itself says robots.txt may not apply to user-initiated fetches by ChatGPT-User.
- Google's AdSense crawler, Mediapartners-Google, now sits in the blocked group too. In April it was explicitly allowed.
Translation: the file is mostly a negotiating position. If you want Transfermarkt numbers in AI answers, ask and pay.
What I'd do now
- Check which of your GEO sources live on platforms like this. If ChatGPT describes your category through a portal that's closing up, your visibility changes without you lifting a finger. Our AI Agent Radar tracks 500 sites.
- If you own unique data (prices, rankings, benchmarks), choose deliberately between visibility and licensing. For most brands, in my view, being readable still beats blocking. The playbook is in our guide to GEO in practice.
- If you block, block cleanly: one list, one goal, and check first that you aren't locking out ad or search crawlers by accident.
Bottom lineIf your data is what AI answers need, you no longer block to disappear. You block to name your price.
Got AI news? Submit news →
FAQ
Does Transfermarkt block ChatGPT?
Its robots.txt blocks GPTBot and ChatGPT-User, but not OAI-SearchBot, the crawler behind ChatGPT search. An RSL licence also prohibits all AI use. Both are requests, not technical barriers.
What is RSL?
Really Simple Licensing is an open standard (version 1.0 since 10 Dec 2025) that lets sites declare machine-readable licence terms for AI use in robots.txt, such as bans, payment or attribution.
When did the new rule appear?
Our AI Agent Radar first recorded the version with the License line on 17 Nov 2026. The Internet Archive copy from 28 April 2026 didn't have it yet.
Source
Transfermarkt: robots.txt (Stand 17.11.2026)
More sources
- Transfermarkt – content url='/', prohibits type='usage': ai-all
- Internet Archive – Ohne License-Zeile; GPTBot, ChatGPT-User, ClaudeBot, Claude-SearchBot bereits gesperrt; Mediapartners-Google ausdrücklich erlaubt
- RSL Collective – Veröffentlicht 10.12.2025; ai-all = jede KI-Nutzung inkl. Training, Inferenz, Indexierung, Retrieval, Grounding; License-Zeile außerhalb einer Gruppe gilt für alle Clients
- OpenAI – 13.12.2023: Zusammenfassungen u. a. von Bild und Welt in ChatGPT mit Quellenlink, Lizenzvergütung
- Wikipedia – Axel Springer seit 2008 Mehrheitseigner (51 %)
- OpenAI – robots.txt gilt für nutzergesteuerte ChatGPT-User-Abrufe möglicherweise nicht
- I4U KI-Agent-Radar – Eigene Daten: neue Fassung erstmals erfasst am 17.11.2026 03:32 UTC




