boxby Simular
DownloadDocsSign inGet a box

Privacy Policy

Last updated: November 19, 2026

This Privacy Policy explains how Simular, Inc. (“Simular,” “we,” “us,” or “our”) collects, uses, discloses, and protects personal information when you visit our websites or access or use our software, cloud services, AI agents, browser and computer-use tools, APIs, integrations, hosted cloud computers and persistent workspaces (including “Simular Box” or “Box”), remote desktop and shell features, and Model Context Protocol (“MCP”) connections, and related services (collectively, the “Services”).

This Policy is written for website visitors, individual account holders, and people who use the Services as authorized users of a business or other organization. When an organization uses the Services and Simular processes personal information on that organization’s behalf, the organization may determine the purposes and means of that processing, and its own privacy notice and contractual terms may also apply. Where you connect an independently operated AI agent or application to Box, that third party may separately determine its own data practices. This Policy describes Simular’s handling of personal information through the Services and does not replace the privacy terms of third-party AI providers that you choose to connect. Whether Simular processes particular information for its own purposes or on behalf of an organization depends on the processing activity and applicable agreements.

1. Information We Collect

A. Information You Provide

  • Account and profile information, such as your name, email address, company or organization, login and authentication information, and account preferences.
  • Prompts, instructions, commands, workflow parameters, messages, and other information you submit when directing the Services.
  • Files and content you upload or make available to the Services, including documents, spreadsheets, images, records, and other data.
  • Support and communications information when you contact us, request support, participate in troubleshooting, or otherwise communicate with Simular.
  • Billing and transaction information associated with a paid subscription. Payment card details may be collected and processed directly by our payment processors rather than stored by Simular.
  • Connected-agent credentials and authorization information. For Box, this may include authentication tokens (stored in hashed form), connector tokens and stored API keys (stored in encrypted form), OAuth client/application identifiers, the user's approved Box resources, the grant's scope and expiration, grant or key creation and revocation events, the identifier of the OAuth grant or personal key used, and associated client IP addresses. A personal access key does not necessarily identify which independently operated AI application is using it.
  • Credit, referral, and promotional information, such as credit grants, balances, and usage; the invitation or promotional link through which you signed up; whether a referral or promotion qualified for credits; and related eligibility and anti-abuse checks.

B. Information Collected or Stored During Use

  • Session and interaction content. The Services may store user inputs, model outputs, conversation history, workflow history, and related tool outputs as part of operating and maintaining the Services. In the main Sai application, text interactions and model responses may be written to session history regardless of whether an optional product-improvement setting is enabled.
  • Screen and interface observations. To perform requested tasks, the Services may read accessibility trees, document object model (DOM) structures, and capture screen or interface observations needed to understand and interact with software interfaces.
  • Execution and diagnostic information, such as commands, tool outputs, task or tool status, timestamps, duration, error information, system events, and resource usage, depending on the Service. For Box, server request logs record the requested tool name, user and Box identifiers, client IP, request status, and duration; they do not store the request's arguments or tool outputs. Shell commands and browser history may nevertheless remain in the Box itself and in saved copies of the Box.
  • Device, browser, network, activity, and metadata, such as IP address, operating system, browser type, device identifiers, features used, timing, and associated technical information.
  • Cookies and similar technologies. Our websites may use cookies, analytics tools, and similar technologies to enable site functionality, maintain security, remember settings, and understand usage. Any optional advertising or cross-context behavioral advertising technologies, and applicable privacy choices, depend on the technologies actually deployed and the requirements of applicable law
  • Box workspace data. When you use Box, content and technical state in the hosted computer may include files you create or download, browser history or website content, shell command history, application settings, and other data stored in the workspace. Browser cookies, saved sessions, credentials, caches, and other application state may also be retained within the Box environment.
  • MCP and connected-agent activity. When a third-party AI application connects to Box, the Services process connection events, tool requests (including commands), and status information, and return task responses to the connected client. Depending on the tools used, such responses may contain screenshots, webpage content, command outputs, file information, or other content from the Box. Simular's Box server request logs capture request metadata, not the contents of MCP arguments or outputs. Where connector tools are separately offered, a connector log records calls and outcomes and stores argument hashes rather than argument content.
  • Security and abuse-prevention records. For Box, the records currently retained centrally for security and operations include account/Box audit events (such as creation, deletion, hibernation, saves, restores, sharing and authorization changes), limited server request metadata, and, where available, connector-call events and outcomes. We do not currently record centrally shell commands, per-file access, running processes, or outbound network destinations inside a Box. Such activity or content may nevertheless be present in workspace histories, files, or saved Box images, which may be reviewed where reasonably necessary as described in Sections 8 and 10.

C. Information from Connected Services

If you connect or authorize a third-party account or service, such as Google Workspace, Slack, GitHub, Linear, Calendly, a practice-management system, or another application, the Services may access, retrieve, transmit, or update information within the permissions you grant and as necessary to perform the actions you request. If you authorize an external AI application to connect to Box through MCP, the application may send task requests to our Services and receive results from the workspace. Such transfers depend on the connection and the tools invoked, and the independent AI provider’s separate privacy terms may apply to its subsequent processing. Data viewed through a browser interface may be different from data received through an application provider’s official API.

2. How We Use Information

We use personal information to:

  • provide, operate, maintain, personalize, and support the Services;
  • execute tasks, workflows, browser navigation, software interactions, and other actions you request;
  • authenticate users and administer accounts and subscriptions;
  • administer usage credits and referral and promotional programs, including verifying eligibility and preventing duplicate, self-referral, or other abusive claims;
  • communicate about tasks, service updates, security notices, billing, and support;
  • detect, investigate, prevent, and respond to security incidents, fraud, abuse, unauthorized access, malicious automation, and attempted access to other users’ workspaces or systems;
  • authenticate users and connected applications, and manage, restrict, suspend, or revoke OAuth grants, personal keys, connected applications, or account access to protect users and the Services;
  • debug, troubleshoot, evaluate, and improve the reliability and quality of the Services, subject to the model-training commitments below;
  • comply with legal obligations and enforce our agreements and policies; and
  • create de-identified or aggregated information that is not reasonably capable of identifying an individual, where permitted by law.

3. AI Processing, Storage, and Model Improvement

Storage of session content is separate from whether content is used to train or improve generalized AI models. The Services may retain session or workflow content as described in this Policy to operate, support, secure, and maintain the Services even when an optional product-improvement setting is disabled.

To perform requested tasks, the Services may send relevant task context to AI model providers or other service providers acting for Simular. Depending on the workflow, this context may include prompts, files, tool outputs, or screen observations necessary to complete the task.

For Box, a third-party AI application selected and connected by you may receive data returned through its MCP calls, which the third party may process under its own terms and privacy policy. The third-party AI provider is not automatically a service provider or subprocessor engaged by Simular merely because you connect it. Simular remains responsible for its own processing and for providers it independently engages, as required by applicable law.

We do not use Customer Content, task transcripts, uploaded files, or screen observations to train or fine-tune public, third-party, or generalized cross-customer AI models unless the applicable customer expressly opts in or otherwise authorizes that use in writing. Certain optional settings may control whether screenshots and associated trajectory data are used for product or model improvement; the scope of those settings may vary by product or deployment.

We do not use Protected Health Information (“PHI”) for generalized or cross-customer model training unless expressly authorized in writing under an applicable Business Associate Agreement and permitted by law.

4. Service Providers, AI Providers, and Subprocessors

We use service providers and subprocessors to help operate the Services, including cloud infrastructure providers, AI model providers, analytics and security vendors, support tools, payment processors, and other vendors. Relevant task context may be transmitted to those providers as necessary to perform requested tasks or support the Services. Third-party processing of task context does not, by itself, mean that the information is used to train the provider’s generalized models.

Providers engaged by Simular to host, secure, or operate Box may receive information necessary for those functions under arrangements appropriate to their roles. By contrast, an AI application you independently connect to Box may receive content at your direction and is governed by its own practices; it is not classified as Simular’s subprocessor solely on that basis.

Where the Services process PHI under a Business Associate Agreement, Simular will use subprocessors for PHI only as permitted by that agreement and applicable law. Simular may maintain a current list of relevant subprocessors on a dedicated subprocessor page or make that list available to enterprise customers on request.

5. Customer-Directed Integrations

A third-party application, account, or service that you or your organization independently selects, contracts for, and directs the Services to access is not necessarily a Simular subprocessor merely because the Services interact with it. Your use of those third-party services is governed by your relationship with the third party, including its terms and privacy practices. You are responsible for reviewing and managing the permissions you grant to connected services and for disconnecting or limiting access when appropriate.

Connecting AI applications to Box. A new OAuth-connected application requires user consent and can access the Boxes selected on the consent screen, which may preselect all Boxes. OAuth grants use the permissions and expiration shown at consent. A personal access key can access all Boxes on the account until revoked and does not require a separate per-agent approval. Box access generally includes available screenshot, browser, file, and shell capabilities; it is not divided into independently adjustable permissions for each such function. Revoking access stops new authorized requests but does not necessarily stop a process the client previously started within Box. Box does not notify you when an additional AI application connects to an existing Box. Some AI applications may register themselves through dynamic client registration; such applications are identified as unverified on the consent screen, and Simular does not verify the identity, security, or data practices of any AI application you connect.

Persistent workspace and multiple connections. Box preserves files and workspace state, including browser sign-ins, cookies, SSH keys, and shell history, through idle periods and reconnections. Multiple authorized AI clients may use the same Box and may access sign-ins or other data left by earlier clients. Disconnecting all agents and credentials leaves existing Box files and website sign-ins intact. There is no separate Box reset option; deleting a Box is the available full reset. Box may automatically hibernate after a period without activity and may subsequently be deleted under the payment-history and subscription-status rules described in Section 9.

User-directed operations and third-party websites. An authorized AI client may direct Box to open websites, interact with accounts, execute shell commands, and create or modify content, subject to applicable permissions and controls. Such actions can expose information visible within the workspace or connected third-party service. You should use only agents and integrations you trust and review their separate privacy practices. Simular does not control an independently selected AI provider’s use of data after that provider receives it, but this does not limit Simular’s own obligations with respect to the Services.

Independent AI providers may have privacy practices that differ from this Policy or from commitments made by your organization. Before connecting an AI application or making information available to it through Box, consider the provider's applicable data-use, model-training, retention, confidentiality, and onward-disclosure terms. This Policy does not change that independent provider's terms or displace Simular's own obligations under applicable law.

6. Google API User Data

Simular’s use and transfer of information received from Google APIs will comply with the Google API Services User Data Policy, including the Limited Use requirements.

  • We request access only to Google API scopes needed to provide the functionality you enable or to perform tasks you request.
  • We use Google user data only to provide or improve user-facing functionality that is prominent and related to the purpose for which access was granted, and as otherwise permitted by the Google API Services User Data Policy.
  • We do not use Google user data for advertising or to build advertising profiles.
  • We do not use Google user data to train generalized AI or machine-learning models.
  • Human access to Google user data is limited to circumstances permitted by the Google API Services User Data Policy.

The commitments above apply to information received through Google APIs. Information accessed through a Google website in a Box browser may involve a different technical access path, although applicable privacy, confidentiality, and security obligations continue to apply.

7. Healthcare and PHI

The general-purpose Services are not intended for processing PHI. Simular may designate specific deployments or service configurations for use with PHI. A healthcare customer may submit or permit access to PHI only through a deployment or configuration that Simular has expressly designated in writing for that purpose and only under an executed Business Associate Agreement (“BAA”). A designation applies only to the identified deployment or configuration and does not extend to other Services.

Certain designated healthcare deployments are intended for administrative healthcare workflows and may use local execution on a customer-controlled device together with cloud or AI processing as needed to perform the requested workflow. The specific handling of PHI, including permitted uses, subprocessors, safeguards, retention, and return or destruction, is governed by the applicable BAA and service configuration.

The Services are not intended to provide diagnosis, treatment recommendations, clinical decision support, or other outputs to be relied upon for a clinical determination unless Simular expressly agrees to such use in writing under terms specifically addressing that use.

8. Human Access

Simular personnel may access Customer Content where reasonably necessary to provide support, investigate security or abuse, maintain or operate the Services, comply with law, or otherwise carry out authorized business functions. The availability and nature of access controls, monitoring, and review may vary by system and deployment. Where additional access restrictions are required by an enterprise agreement, data processing agreement, or BAA, those terms will govern within their scope.

For Box, there is no ordinary in-product staff viewer for a customer's workspace, and Simular personnel do not routinely open users' Boxes. However, authorized engineers with infrastructure administration access may be able to reach Box content and saves, and may do so where reasonably necessary for security, abuse investigation, support, operations, or legal compliance. Such access is subject to internal access restrictions and applicable law.

9. Data Retention and Deletion

Retention periods vary by data type, product, workflow, deployment, operational need, and applicable legal or contractual requirements. Some execution environments or temporary workspaces are designed to be short-lived, while session history, workflow history, generated deliverables, logs, backups, security records, or other information may be retained for longer periods.

Box-specific retention and inactivity deletion. Box may hibernate after a period without activity. Boxes associated with active Small or Medium subscriptions are not automatically deleted solely for inactivity while the subscription remains active. If an account has ever made a payment for a top-up or subscription, its Box is subject to automatic deletion, together with all associated saves, after 30 consecutive days in hibernation when no paid subscription is active. If the account has never paid, the corresponding period is 14 consecutive days in hibernation. If a paid subscription is cancelled, the applicable inactivity-deletion period begins no earlier than the end of the paid subscription period; after a failed payment, it begins no earlier than when payment retries have been exhausted and the subscription actually ends. The applicable deletion rules are based on activity, payment history, and subscription status, not remaining credit balance. For Boxes existing when this Policy is posted on November 19, 2026, the applicable inactivity period starts no earlier than that date, and no such Box will be deleted for inactivity earlier than 14 days after that date. We will send a warning email to the email address associated with the account at least seven days before any inactivity deletion. If the warning was not sent at least seven days before a planned deletion, the deletion will be delayed until at least seven days after it is sent. We may also send a reminder approximately one day before and a notification after deletion. You are responsible for keeping your account email address current; non-receipt of a sent email does not itself prevent deletion. We may change these inactivity rules prospectively by updating this Policy and providing any notice required by law. Revoking a connected application's authorization or deleting a conversation with that application does not delete information already stored in Box.

Box saves and recovery. Box may be saved automatically approximately every 15 minutes and through user-initiated saves. Box saves and archived Box data may be stored through different infrastructure providers, including Google Cloud Storage and Cloudflare R2. Archived Boxes previously stored on bare-metal servers may be transferred to Cloudflare R2. Each user's saved and archived Box data remains encrypted using a user-specific key. While a Box exists, Simular generally retains all saves made during the preceding 24 hours, one daily save for 30 days, and one weekly save for as long as the Box exists. Deleting a Box deletes its active environment and associated saves from ordinary use, although copies may remain recoverable from storage for up to seven days after deletion.

Operational log retention. For Box, account/Box audit events are generally retained for 180 days, connector-call logs for 90 days where applicable, and limited server request logs for 30 days. Audit events may remain for up to 180 days following account closure. Shell history and browser history stored inside the Box follow the Box and its saves rather than these server log periods.

Security record preservation. Following suspected misuse or a security incident, Simular may preserve relevant authentication, account/Box audit, connector, and server request records and, where reasonably necessary, affected Box content or saves, notwithstanding Box deletion (including inactivity deletion), account closure, or the retention periods otherwise described in this Policy, as reasonably necessary to investigate or remediate the incident, prevent recurrence, comply with law, or establish, exercise, or defend legal claims. Preserved information remains subject to access restrictions and security protections and will be deleted when no longer needed for those purposes.

You may request deletion of personal information as described below. Deletion may not be immediate and may not remove information from backups, archives, security logs, or systems from which immediate deletion is not technically practicable. Where deletion is required, information will remain subject to applicable protections until deletion occurs through the relevant lifecycle. Enterprise agreements or a BAA may establish additional retention or deletion requirements.

To delete a Box or request account closure, use any applicable product controls or contact support@simular.ai. Disconnecting an AI application, revoking a key, deleting a Box, and closing an account are separate actions. A deleted Box and its saves may remain recoverable in backup storage for up to seven additional days. Users cannot independently delete individual Box saves. Account closure is handled through support: the account is disabled promptly, related Boxes are deleted and connector grants revoked, and account data is scheduled to be erased within 30 days, subject to legally required or otherwise permitted limited retention, including security audit records and anti-abuse identifiers described below.

Post-deletion anti-abuse information. To prevent fraudulent, duplicate, or abusive claims of Free sign-up, referral, or promotional credits, we may retain limited anti-abuse identifiers for up to three years after account deletion, subject to applicable law and earlier deletion where no longer necessary. Such identifiers may include hashed or otherwise pseudonymized information derived from account identifiers, IP addresses, and third-party authentication accounts, as well as payment-method identifiers or fingerprints provided by payment processors. We use this information for fraud and abuse prevention, not for advertising or general account profiling. Such information may constitute personal information and remains subject to applicable privacy laws and individual rights, including applicable deletion rights.

10. Security

We maintain administrative, technical, and organizational safeguards designed to protect personal information against unauthorized access, use, disclosure, alteration, and destruction. Our security and compliance program is designed to evolve with the Services, the sensitivity of the data we process, and applicable contractual and legal obligations. No method of transmission, storage, or processing is completely secure, and we cannot guarantee absolute security.

For cloud execution environments such as Box, we maintain safeguards designed to reduce unauthorized access between customer accounts and hosted workspaces. Depending on the deployment and technical configuration, measures may include logical access controls, server-side authorization checks for workspace requests, and controls intended to separate computing, storage, or network access. The particular measures in use should be understood in light of the Service configuration; no method of processing or storage is completely secure.

Box security logging and response. To prevent abuse, investigate suspected incidents, and operate the Services, Simular records certain account and Box lifecycle/audit events, connector-call records where connectors are offered, and limited server request metadata. The Box audit log identifies relevant actors, credentials or grants, IP addresses, actions, and outcomes, and is generally kept for 180 days. Connector logs record call results and hashes of arguments rather than the arguments themselves, are visible to the user, and are generally kept for 90 days. Box server request logs contain tool names, user/Box identifiers, IP addresses, statuses, and durations without tool arguments or responses, and are generally kept for 30 days. We do not currently collect centrally shell-command content, per-file access, process activity, or outbound network destinations within a Box. Simular may use the records it retains to detect, investigate, and respond to suspicious activity and to restrict a grant, key, or account. Where reasonably necessary to investigate or mitigate a suspected threat, authorized personnel may also review relevant Box content, such as files, screenshots, shell history, or browser state, subject to the restrictions described in Section 8. We may use relevant information to investigate threats, enforce our Terms, preserve evidence, and take protective measures. Stopping already-running processes or a Box's outbound network traffic may require manual infrastructure intervention. Security monitoring is not continuous and cannot guarantee detection of every harmful action.

For Box, a personal access key may identify a credential and client IP address without identifying the AI provider using it. Revoking that key may therefore affect multiple applications that share it. OAuth grants and personal keys may be revoked, but protective restrictions cannot necessarily be applied to each individual running tool or session.

Credentials and secrets. Box access tokens are stored in hashed form, and connector tokens and stored API keys are encrypted. However, content inside a Box, including files, shell history, browser state, and saves, may contain passwords, keys, or other secrets entered by you or a connected AI application; such content is not separately filtered or redacted and will persist in the Box and its saves until the Box is deleted.

We may use independent assessments, control frameworks, monitoring tools, and compliance-management systems as part of our security program. References to a framework or compliance program do not mean that Simular has obtained a certification or attestation unless we expressly state that the certification or attestation has been completed.

11. How We Disclose Information

We may disclose personal information:

  • to service providers and subprocessors that perform services for us;
  • to third-party services you direct us to connect with or interact with;
  • to third-party AI applications that you authorize to connect to Box, through responses to their MCP tool requests, which may include workspace observations, tool outputs, or other task-related content;
  • to comply with law, legal process, or valid governmental requests, or to protect rights, safety, and security;
  • in connection with a merger, financing, acquisition, reorganization, bankruptcy, or sale of all or part of our business, subject to appropriate protections; and
  • with your consent or at your direction.

We do not sell Customer Content. We do not use Customer Content or Google user data for targeted advertising. If our use of website analytics, cookies, or similar technologies is treated as a “sale” or “sharing” of personal information under applicable law, we will provide any required disclosures and choices.

12. Your Privacy Rights

Depending on where you live and subject to applicable exceptions, you may have rights to request access to, correction of, deletion of, or portability of personal information; to object to or restrict certain processing; to withdraw consent where processing is based on consent; and to receive information about certain disclosures or processing activities. You may also have the right not to be discriminated against for exercising applicable privacy rights.

To submit a privacy request, contact support@simular.ai. We may need to verify your identity or authority before completing a request. If you use the Services through an organization and Simular processes your information on that organization’s behalf, we may direct your request to that organization or assist it in responding.

If you connect an independent AI application, requests relating to data that the third-party AI provider separately receives or retains may need to be directed to that provider. A request to Simular concerning data stored in Box is not the same as a request to delete data held in an external AI application. Where applicable U.S. state privacy law provides additional rights or choices, we will honor those rights as required by law, subject to applicable exemptions and verification procedures.

13. International Data Transfers

Simular is based in the United States, and personal information may be processed in the United States and other countries where Simular or its service providers operate. Where required by applicable law, we use appropriate transfer mechanisms and safeguards for cross-border transfers, which may include contractual protections or other legally recognized mechanisms.

14. Children

The Services are not directed to children under 18, and we do not knowingly collect personal information from children under 18 through the Services. If you believe a child has provided personal information to us in violation of this Policy, please contact us.

15. Changes to This Policy

We may update this Privacy Policy from time to time. If we make material changes, we will provide notice through the Services, by email, or by another reasonable method before the changes take effect when required by law. The “Last updated” date above indicates when this Policy was most recently revised.

Contact Us

Simular, Inc.

1209 Orange Street

Wilmington, New Castle, Delaware 19801

Email: support@simular.ai

box by Simular
Sign inDownloadDocsBlogAccessSimularTermsPrivacy