# Does my AI company see my data?

Checked 2026-11-14 against GilbApp 49467c2b74111a7021a0419594f5f64ec3cc0eb0.

Yes: what you type and what your AI reads from your vaults goes to Anthropic or xAI, under your own agreement with them.

## Deeper

Also asked:

Does Anthropic or xAI train its AI on what my AI reads in my vault? Anthropic does, for Free, Pro, Max and Ultra accounts while its "Model Improvement" setting is on, and not under its commercial terms unless you chose it. xAI's help page does not say; xAI's privacy policy, https://x.ai/legal/privacy-policy, sets it.

How do I stop my AI company from using my data for training? For Claude, turn off "Model Improvement" at claude.ai/settings/data-privacy-controls. For Grok, xAI's help page sends you to "Profile → Settings → Data Controls" for ways to control your data; it does not say whether training is one of them.

Your AI company is the company that makes the AI you connected: Anthropic for Claude Code, xAI for Grok. Both are US companies.

MyJackfruit's Privacy page, last updated November 11, 2026, says: "It is your AI, not ours: what you type to it, and what it reads from your vaults, goes to its provider under your own agreement with them. We do not send your data to an AI on your behalf." \(https://myjackfruit.com/privacy/\). MyJackfruit is the public brand, at myjackfruit.com; it is not another name for GilbPlatform. Its Terms say that your use of the AI you connect "is also governed by that provider's terms" \(https://myjackfruit.com/terms/\).

So your AI company gets what you type and what your AI reads, not your whole vault. In GilbApp's chat the agent works as you, so it can read any vault you can reach; in Terminal, Claude Code reaches only the vaults you gave its AI vault access to. To keep a vault away from your AI, do not give its AI vault access there. In GilbApp's chat the agent works in the vaults the middle of the screen lists, and in any vault you name, so check that list before you send a message.

With Grok, GilbPlatform's server passes the chat, with what the agent reads, on to xAI, using your own SuperGrok sign-in. What xAI does with it is in xAI's privacy policy, https://x.ai/legal/privacy-policy. Grok's help page says that for ways to control your data you go to "Profile → Settings → Data Controls" \(https://grok.com/faq, read 13 November 2026\).

With Claude Code, Anthropic says: "Claude Code runs locally. To interact with the LLM, Claude Code sends data over the network. This data includes all user prompts and model outputs, encrypted in transit via TLS 1.2+." For Free, Pro, Max and Ultra accounts it says: "We will train new models using data from Free, Pro, Max, and Ultra accounts when this setting is on." You change that setting at claude.ai/settings/data-privacy-controls. For those accounts, Anthropic keeps your data for 5 years when you allow it to be used to improve its models, and for 30 days when you do not. Anthropic's page on model training says it will use your chats and coding sessions in some cases, one of them: "Your conversations are flagged for safety review \(in which case we may use or analyze them to improve our ability to detect and enforce our Usage Policy, including training models for use by our Safeguards team, consistent with Anthropic's safety mission\)" \(https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training, read 13 November 2026\). For accounts under Anthropic's commercial terms it says: "Anthropic does not train generative models using code or prompts sent to Claude Code under commercial terms, unless the customer has chosen to provide their data to us for model improvement." \(https://code.claude.com/docs/en/data-usage, read 13 November 2026\). Anthropic calls the setting "Model Improvement" \(https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training, read 13 November 2026\).

Anthropic and xAI set these rules and may change them; their own pages are the ones that count.

This answer is written for a computer screen.

Source: MyJackfruit Privacy \(https://myjackfruit.com/privacy/, "Last updated: November 11, 2026", read 12 November 2026\), section "When you connect an AI."; Terms \(https://myjackfruit.com/terms/, same date\) section "AI". GilbPlatform 670005cc \(https://id.gilb.com/version read 12 November 2026\): src/ai/complete-chat.ts:59 \(the chat's model calls go to https://api.x.ai\), src/ai/xai-oauth.ts:1-18 \(the person's own SuperGrok sign-in\), src/routes/ai-chat.ts:86 \(messages\). Live answers reused: "How do I chat with an agent in GilbApp?" \(works as you\), "How do I give my AI access to a vault?" \(reach\), "What does it cost for an AI?" \(MyJackfruit brand, provider's terms\). Makers' pages read 13 November 2026: https://code.claude.com/docs/en/data-usage; https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training \(last updated March 16, 2026\); https://grok.com/faq. xAI's privacy policy refused automated reading \(HTTP 403\) and is named, not quoted.
