The design
Grok is the conversational AI built by SpaceXAI (the company formerly known as xAI, now folded together with SpaceX under that name), and it includes companion-styled personas, Ani and Rudy, alongside its general-purpose chat mode. The company's own Privacy Policy, effective 24 August 2026 as retrieved on 16 September 2026, governs the Grok app and Grok.com. It asks that users 'do NOT include personal information' in prompts, while acknowledging it 'cannot control what you provide.' The policy describes one unified set of rules for 'personal information we collect' from anyone using the Service.
What the evidence says
Reading the full document, neither 'Ani' nor 'Rudy' nor the word 'companion' appears anywhere in the Privacy Policy. It does acknowledge that, 'depending on how a user interacts with the Service, including which modes the user purposely selects,' content may include 'suggestive dialogue... or sexual situations,' an indirect reference to companion-style modes without naming them or carving out separate data rules for them. On retention, the policy states that with Private Chat turned on, conversations 'will be deleted from SpaceXAI systems within 30 days,' and the same 30-day figure applies if a user deletes a conversation or account, each subject to exceptions for 'legal, compliance, or safety purposes.' These are the company's own stated practices; no independent audit of Ani or Rudy conversation handling specifically appears in this document.
What it asks of people
The company's own Terms of Service confirm the same pattern: anyone talking to Ani or Rudy is, by the policy's own silence on the subject, agreeing to the same data terms as a person asking Grok to summarize a document or write code. There is no companion-specific consent screen or separate disclosure describing how conversations with a persona designed to be affectionate are handled differently, because the policy does not treat them differently. A user who wants the 30-day deletion figure to apply must actively enable Private Chat or delete conversations rather than relying on a default.
Privacy and safeguards
What is disclosed: a uniform collection-and-retention framework, a 30-day deletion window tied to specific user actions, and an acknowledgement that some 'modes' may produce sexual or suggestive content. What is not disclosed: any distinct retention, training-data, or age-gating rule specific to companion personas, or whether conversations with Ani or Rudy are treated as more sensitive internally despite the uniform policy text. No regulator's review of this treatment is described in the document.
- Should a companion persona designed for emotional engagement carry different data rules than a coding assistant, and does this policy's silence answer that question?
- Is Private Chat mode on by default, and would a typical user know to enable it before an intimate conversation?
- What happens to a companion conversation's data during the 'legal, compliance, or safety' exceptions to the 30-day window?
The absence of companion-specific language in a general privacy policy is itself a finding: it means a conversation with a persona built to feel personal is, as a matter of documented policy, handled exactly like any other prompt.
Sources & reading trail
Shows the policy applies uniformly to Grok including companion modes, states the 30-day Private Chat deletion window, and acknowledges suggestive content in some modes.
Source published: Not established · Retrieved: 16 September 2026
Confirms the same absence of companion-specific (Ani/Rudy) language and states the general minimum-age terms that apply to all Grok use.
Source published: Not established · Retrieved: 16 September 2026
Product documents, regulator records and studies establish the entry; the design reading is AI Companions editorial analysis. This retrospective draft does not imply the site published on the event date.