Privacy Policy

Last updated: October 4, 2026

CLIO is an open-source AI workspace developed by the Gnosis Research Center at Illinois Tech as part of IOWarp. This policy explains how the CLIO software and the website at clio.iowarp.ai handle information.

CLIO runs on a computer or server that you choose. If your institution or another organization operates that server, its administrator controls that installation and may have access to its stored data. Ask that operator about its own privacy, access, retention, and backup practices. Installing CLIO does not automatically send your workspace to the CLIO project.

Information CLIO handles

Your work: messages, attachments, selected files and folders, generated results, session history, memory, and records of the actions taken during a run.
Connected sources: file and folder names, paths, identifiers, sizes, revisions, and the contents you ask CLIO to read, download, or analyze.
Connection information: server addresses, source settings, and authorization tokens needed to access the services you connect.
Support information: information you choose to include in a support request, diagnostic report, or issue. Public GitHub issues are visible to others; do not include private files, passwords, or tokens in them.

CLIO uses this information to provide the features you request: browsing and linking data, transferring files, running analyses, producing results, and keeping the supporting evidence available in your workspace.

Google Drive and Globus

Connecting an account starts a sign-in and permission request with the provider. You enter your Google or Globus account credentials on that provider’s sign-in page. CLIO receives authorization tokens so it can continue the connection without asking you to sign in for every file.

For Google Drive, CLIO accesses metadata and file contents for the folder or files you select. It can download ordinary files and export Google documents to formats it can read. Read-only connections request permission to view and download Drive files. Connections configured for supported changes can request broader Drive access. Google’s permission grant can cover more of your Drive than the folder selected in CLIO; the selection is an application-level limit, not a separate Google permission boundary. Review the permissions shown by Google.

For Globus, CLIO uses your permission to browse the collections you select, read selected files, and submit transfer jobs when you request a download and receiving storage is available. Globus and the collection operators process the requests and transfers under their own policies.

When data leaves the CLIO machine

When you ask an agent to work with data, relevant messages, file contents, and tool results can be sent to the model provider configured for that conversation. That provider may be on your computer, on another server, or a cloud service. Attaching a folder tells the agent the folder exists; reading files from it can then include their contents in model requests. Check the chosen provider’s data handling and retention terms before using sensitive information.

Tools, extensions, connected storage services, and remote execution hosts can receive the data needed for the actions you request. Sharing a session, exporting results, or uploading a diagnostic report can also share its contents with the destination you select. A remote CLIO server and a remote model are separate destinations, even when you use the interface in your own browser.

The CLIO project does not sell connected-source data or use it for advertising. It does not collect Google user data to train generalized AI or machine-learning models. Use model providers and settings that do not use Google data for that purpose. Google data is used for the user-facing features you authorize. Project maintainers do not have routine access to files in your installation; any human access to Google data shared with the project must be limited to your affirmative permission or the security and legal exceptions permitted by Google.

CLIO’s use and transfer of information received from Google APIs will adhere to the Google API Services User Data Policy, including its Limited Use requirements.

Storage and security

Workspace files, downloaded copies, cached linked files, messages, and retained evidence are stored on the machine running CLIO. Linking a folder keeps the original files at their source and can cache a file locally when it is opened or read. A download creates a separate copy on the CLIO machine.

Saved Google and Globus tokens are stored separately from source descriptions and agent-visible results. Browser SFTP passwords and uploaded private keys are kept in memory for the active connection; host-configured credentials follow the host’s configuration. Access to stored information depends on that machine’s accounts, file permissions, and security settings. Operators are responsible for securing the host, its backups, and remote connections. This policy does not promise that workspace files or token storage are encrypted at rest.

Your choices and deletion

Disconnect a source to stop CLIO using that folder connection. Your saved Google or Globus sign-in remains available for other sources and workspaces on the same CLIO. Downloaded files and evidence already used in a message remain.
Sign out of a provider to remove its saved sign-ins from this CLIO across your workspaces. This does not delete downloaded files or remote originals.
Revoke provider access in your Google Account connections or Globus application permissions. Disconnecting in CLIO does not itself revoke every provider-side grant or disconnect another installation.
Unlink a folder to remove its workspace link. Remove downloaded copies separately when you no longer need them. These actions do not delete originals in Google Drive, Globus, or another source.
Remove an unsent attachment to withdraw the temporary copy or link it added. Data already retained by a sent message or another attachment is preserved.
Delete retained information separately. Session history, extracted text, memory, results, downloaded copies, exports, and backups can have separate lifetimes. Removing a connection or message does not erase all of those copies. See data locations and uninstall and data removal, or ask the operator of your CLIO server for help.

CLIO has no single automatic retention period for all workspace data. Retained data remains until you or the installation’s operator remove it. Providers and recipients may retain information already sent to them under their own policies; local deletion does not delete those copies.

Website visits and routine network requests

The public website is hosted on GitHub Pages. The host processes connection information, including IP addresses and request information, as described in GitHub’s Privacy Statement. The site can save interface preferences in your browser. It does not include advertising trackers or a project analytics service.

CLIO may contact PyPI or GitHub to check for updates and download public model metadata from models.dev. Those services receive ordinary connection information. These checks do not require your conversation or connected-source file contents. CLIO does not include a product-usage telemetry service. See the Data and network guide for technical details and update settings.

Contact and changes

For information stored in an institution’s or another operator’s installation, contact that operator about access, correction, export, or deletion requests. For project privacy questions, the private contact address will be added before this policy is published. Never post sensitive information in a public issue.

Updates to this policy will appear here with a revised date. A new use of Google data outside the permissions and purposes already disclosed requires a new disclosure and consent before that use begins.