Can I upload my thesis?
Not just because a tool accepts the file. First check whether you may share the text and data, and whether your university allows that account. Then check training settings and storage. This table brings together the privacy facts we already have. An unknown entry is a reason to ask, not permission to upload.
What happens to the file?
The table covers the ten tools in our unpublished-work guide. Click a column heading to sort its text; click again to reverse the order. Sorting happens in your browser without a network request. It is not a privacy ranking.
Read each account limit. A university email address does not turn a personal account into an education or enterprise account. No training does not mean no storage, no human review or permission to share someone else's work. Check the linked policy and your actual settings before uploading.
- Read the permission and confidentiality guide
- find a tool for unpublished work
- Check your thesis AI-use rules
| ChatGPTDossierSource 1Source 2Source 3Checked on: | Personal Free / Plus / Pro: content may be used to improve models. Temporary Chat is not used to improve models. | Settings > Data controls > Improve the model for everyone: turn off before starting new conversations. Feedback can make the whole associated conversation available for training. | Temporary Chat: not used to improve models; may be kept up to 30 days for safety. General upload retention and human-review limits are not confirmed in this evidence set. | Inputs and outputs are not used to improve models by default. Applies to Business / Enterprise / Edu / API. API: inputs and outputs retained up to 30 days; zero data retention (ZDR) on request for eligible endpoints with a qualifying use case. Do not transfer API limits to Edu or every file-upload feature. |
|---|---|---|---|---|
| ClaudeDossierSource 1Source 2Source 3Source 4Source 5Checked on: , | Personal Free / Pro / Max: users choose whether chats improve Claude. The initial setting is unknown / not checked. | Consumer model-improvement choice is documented; the exact setting path is unknown / not checked. | New or resumed consumer chats: five years if training is allowed, otherwise 30 days. Enterprise offers custom retention. | Team / Enterprise: no training on customer content by default. The consumer choice does not apply to Team / Enterprise / Education / API; Education-specific terms are unknown / not checked. |
| ConsensusDossierSource 1Checked on: , | The privacy policy says user data is not used to train LLMs, including third-party models. Queries and uploads still go to model subprocessors. | The policy states no LLM training; a separate training opt-out is unknown / not checked. | Account settings allow clearing thread history and deleting the account. A retention duration is unknown / not checked; deletion is not a no-storage promise. | Unknown / not checked |
| ElicitDossierSource 1Source 2Checked on: , , | Unknown / not checked | Unknown / not checked | Unknown / not checked | Enterprise: no training on customer data by default. Lower-tier and education-specific terms are unknown / not checked. The privacy policy says its servers are in the United States. |
| GeminiDossierSource 1Source 2Source 3Checked on: , | Consumer: with Keep Activity on, activity is used to improve Google services, including training generative AI models, and human reviewers review some of the data. | Turn Keep Activity off: future chats won't be used to train Google's AI models, unless you send Google feedback. | Auto-delete default 18 months; with Keep Activity off, chats are kept up to 72 hours. Chats reviewed by humans can be kept up to three years. | Workspace (core service): not human-reviewed or used to improve generative AI models; Workspace additional services follow consumer terms. Gemini API: unpaid use may improve Google products, paid use does not; in the EEA, Switzerland and the UK the paid terms apply to all use. Google AI plans are consumer bundles, not Workspace protections. |
| Microsoft CopilotDossierSource 1Checked on: | Unknown / not checked | Unknown / not checked | Microsoft 365 work account: interaction data is processed and stored in line with the organisation's contractual commitments for its other Microsoft 365 content; admins can set retention policies in Microsoft Purview. LLM calls can go to other regions. Duration and personal-account retention are unknown / not checked. | Microsoft Copilot (Microsoft 365): prompts, responses and Graph data are not used to train foundation LLMs. This does not establish the consumer default or education-specific terms. |
| NotebookLM / Gemini Notebook EnterpriseDossierSource 1Source 2Source 3Checked on: , | Unknown / not checked | Unknown / not checked | Unknown / not checked | Gemini Notebook Enterprise (Google Cloud): data stays in your Google Cloud project and can't be shared externally; uploads are subject to the Google Cloud Terms of Service. Human review and model-improvement use: unknown / not checked. Do not apply this to a personal account. Education-specific terms are unknown / not checked. |
| PerplexityDossierSource 1Source 2Source 3Source 4Checked on: , | Consumer searches may be used to improve Perplexity AI by default. File-upload help separately says file content only customises responses. | Account Settings; the help page calls the setting AI Data Retention under Preferences. Applies only to data collected after opting out, not earlier training data. | File-upload help: files stay private, content only customises responses; files retained for 30 days. General human-review limits are not confirmed here. | Enterprise / API: the API docs say no training by default, but certain APIs may be subject to product-specific terms. Per the API docs, zero data retention is not the default; it is available only to eligible customers through an account-level configuration under a written agreement. File-upload help separately says 7 days for Enterprise Pro files; confirm feature scope. Education-specific terms: unknown / not checked. |
| SciSpaceSource 1Source 2Checked on: | Terms: no training of a general-purpose model unless separately disclosed and permitted by applicable rights, Source Terms and law. No training opt-out was found in the checked privacy policy. This is not an unconditional no-training guarantee. | No training opt-out found in the checked privacy policy. This is not proof that none exists. | Unknown / not checked | Unknown / not checked |
| SciteDossierSource 1Checked on: , | The policy says Customer Data is not used to train, fine-tune, retrain or otherwise improve any large language models. | Unknown / not checked | Unknown / not checked | The policy has a FERPA section for students using institution-licensed access. Other education / enterprise differences are unknown / not checked. |
If an answer is missing, wait
Ask your supervisor or university IT which account and data types are approved. Keep interview transcripts, confidential drafts and a manuscript you are reviewing out of a cloud assistant while permission is unresolved. For a question about structure, a short example without the underlying research data may be enough.
Questions this guide answers
Every answer below is assembled from the dated fields on this page. Nothing is written separately for search.
Does a no-training policy mean I can upload?
No. You still need permission to share the work and data, and approval for the account and feature. Training, retention and confidentiality are different questions. A paid plan or a university email address does not answer them.
What does unknown / not checked mean?
The allowed evidence does not establish that fact for that account. It does not mean the vendor has no policy, nor that uploading is safe. Read the current policy and ask your university before sending unpublished material.