In short. A public DeepSeek shared conversation may be obtained by crawlers, and robots rules do not provide access authorization.
Definition: Yes. DeepSeek's Terms of Use says that a shared conversation may be viewed by anyone with the link and, if it is published on public networks, may be obtained by third parties through web crawlers. DeepSeek says its robots exclusion controls reduce scraping, while technical limits leave a residual risk.
Key facts
- DeepSeek calls the combined user inputs and model outputs “Dialogues” and lets users share them through a unique URL, according to its Terms of Use.
- Anyone who has access to a shared link can view the linked Dialogues.
- The Terms of Use identify web crawlers as a technical means by which third parties may obtain Dialogues that are later published on public networks.
- DeepSeek says it has implemented the robots exclusion protocol as an anti-crawling mechanism, but says technical limitations prevent the risk from being entirely eliminated.
- The IETF's Robots Exclusion Protocol says crawlers are requested to honor these rules and describes them as access instructions rather than authorization.
What does DeepSeek document about a shared conversation?
The Terms of Use page is marked “Last Update: March 27, 2026”. It defines a shared item as the user's Inputs and the service's Outputs together, labels that material a Dialogue, and says that sharing creates a unique URL. The same section states that anyone with access to the shared link can view the linked Dialogues.
DeepSeek describes the crawler issue conditionally. The risk applies when a Dialogue shared with others is subsequently published on public networks. The wording documents a possibility of third-party acquisition through technical means such as web crawlers. The terms leave collection of any particular crawler or DeepSeek link unreported.
Why does robots.txt leave a residual risk?
RFC 9309, published in September 2022, describes the Robots Exclusion Protocol as rules that crawlers are requested to honor. It states that those rules are a form of access instruction rather than authorization and says valid content security measures are required alongside the protocol. A public URL therefore remains a public URL even when a site publishes crawler instructions.
The protocol uses a user-agent group and path rules. Its example syntax is:
User-Agent: *
Disallow: *.gif$
Disallow: /example/
Allow: /publications/
RFC 9309 says a crawler matches the path against Allow and Disallow rules and uses the most specific match. It also says crawlers may cache a fetched robots.txt file and should not use the cached version for more than 24 hours unless the file is unreachable. They operate as crawler instructions; confidentiality requires an application-layer control for the shared conversation.
What should be reviewed before public sharing?
DeepSeek's Privacy Policy, marked “Last Update: Feb 10, 2026”, repeats the warning for Dialogues shared through a unique URL. It recommends avoiding your own or another person's personal information, especially sensitive personal data; obtaining explicit consent before sharing another individual's personal information; and reviewing a conversation before publishing it to an open network.
The same policy says DeepSeek may obtain publicly available Personal Data through online sources to train its models and provide Services. That statement concerns publicly available Personal Data generally. The documents reviewed here leave the crawler names, collection timing, search treatment, retention period, and removal process for a particular shared conversation unspecified.
Related terms
- DeepSeek web search
- What does DeepSeek share with external search APIs?
- robots.txt
- robots.txt Disallow directive
- robots.txt caching and error handling
FAQ
Does a shared link make the conversation viewable?
Yes. DeepSeek's Terms of Use says that anyone with access to the shared link can view the linked Dialogues.
Does robots.txt make a public conversation private?
No. RFC 9309 describes robots rules as instructions requested of crawlers and says they are not access authorization. DeepSeek's Terms of Use also says technical limitations prevent the crawler risk from being entirely eliminated.
Does DeepSeek say that every shared conversation is collected?
No. The terms describe a risk for Dialogues published on public networks. They leave collection of every shared conversation and the identity of a crawler for a specific link unreported.
Sources
- DeepSeek, “DeepSeek Terms of Use”, March 27, 2026
- IETF, “Robots Exclusion Protocol”, September 2022
- DeepSeek, “DeepSeek Privacy Policy”, Feb 10, 2026
Check how a page reads to answer engines. Open the GEO score analyzer