DeepSeek · Training
DeepSeekBot
The collection is admitted; the collector is not. DeepSeek's privacy policy states verbatim that it "may obtain publicly available Personal Data via online sources to train our models and provide Services" (and that second clause implies retrieval beyond training), while naming no user agent, no crawler and no opt-out route, and the transparency centre it links covers model cards and algorithm disclosure rather than crawling. Two hard negatives sharpen it: `deepseek.com/robots.txt` returns 404, so a company whose models learn from robots-governed content publishes no robots file for its own site, and `deepseek.com/about` returns 404 as well. That second one matters because it is the exact `+` URL carried by the `DeepSeekBot` string that circulates in blocklists, which means that string's self-identification target does not exist.
- Operated by
- DeepSeek
- Purpose
- Training
- robots.txt token
- DeepSeekBot
- Verification
- User agent only
Collects pages into a corpus used to train models.
How to verify DeepSeekBot
DeepSeek publishes no list of IP addresses for DeepSeekBot, so nobody (us included) can prove that a request carrying this user agent really came from DeepSeek. The name is trivial to copy. Treat it as a claim the visitor is making about itself, not as an identity anyone has checked.
There is no range list to confirm. This entry was last reviewed against DeepSeek's own documentation on .
User agent
DeepSeek has not published a full user-agent string for DeepSeekBot. Requests are identified by the DeepSeekBot product token appearing in the User-Agent header; we match that token rather than a whole string, because the rest of the header varies and matching it would miss real traffic.
robots.txt for DeepSeekBot
Block
User-agent: DeepSeekBot
Disallow: /Allow
User-agent: DeepSeekBot
Allow: /robots.txt is a request, not an enforcement mechanism. It is honoured by convention, and a crawler that ignores it is stopped at your edge, not in a text file.
What blocking DeepSeekBot costs you
Nothing is surrendered and nothing is secured, which is a stranger place to stand than it first sounds. There is no publicly documented live-retrieval crawler and no citation surface a site can be dropped from, so the disallow forfeits no referrals, no link previews and no answer-panel mentions. It also guarantees nothing: with no published token and no stated commitment to robots.txt, `User-agent: DeepSeekBot` may be a string that never matches anything the vendor sends. Anyone who wants an actual assurance here needs network-layer controls, because the directive is a preference filed with a company that has not said it reads them.
Vendor documentation
DeepSeek does not publish documentation for DeepSeekBot that we could find. Everything on this page comes from what they do publish elsewhere and from observed behaviour, so treat it accordingly.