Privacy Policy

Your privacy, by product

This policy distinguishes SquadOS Web, Studio Desktop and SquadOS Mobile. Use this summary to find your situation and read the conditions and limits in the relevant sections.

• SquadOS Web: collaboration and organization data processed in the cloud.
• Studio Desktop: local state and real files; context sent to AI and the services used.
• SquadOS Mobile: cloud collaboration, data on the phone and remote access to an authorized Desktop.
• Organizations: access according to the resource, permissions and sharing.
• Encryption of new conversations: optional per organization, with limits and authorized backend access.
• Desktop telemetry: optional diagnostics enabled by default and switchable off; separate filters for counters and errors.

OpenRouter, Composio, cookies and replay descriptions in the general sections apply to flows using those services. They do not mean that every action in every product uses the same providers. Sections 16 to 24 explain these differences.

1. Who We Are

SquadOS is operated by CSC Labs LLC, a limited liability company organized under the laws of the State of Florida, United States. CSC Labs LLC is the data controller for personal information processed for its direct relationship with users of the SquadOS platform and associated services. Organization content may involve a different controller/processor relationship, as explained in section 19.

Our Data Protection Officer (DPO) / Encarregado de Dados (LGPD) can be reached at [email protected] for all privacy-related inquiries, data subject requests, and compliance questions. We aim to respond to all privacy inquiries within five business days.

Our registered address is 7680 Universal Blvd Ste 380, Orlando FL 32819. All formal legal notices must be sent to this address or to the email above.

As of the date of this policy, our EU representative under GDPR Article 27 is pending appointment. [PENDENTE — to be appointed before EU outreach] We acknowledge this is a known gap and are actively working to appoint an EU representative before commencing active marketing to residents of the European Economic Area, the United Kingdom, or Switzerland. Until that appointment is made and published here, EEA/UK/CH residents who use the Service do so at their own discretion.

2. Scope

This Privacy Policy applies globally to all users of the SquadOS platform, website, edge functions, public agent endpoints, and any other services operated by CSC Labs LLC (collectively, the "Service"). It governs how we collect, use, store, disclose, and protect personal information regardless of the country from which you access the Service.

Brazilian residents (LGPD): If you are located in Brazil, the Lei Geral de Proteção de Dados (Law No. 13,709/2018 — LGPD) applies to our processing of your personal data. The provisions throughout this policy that reference LGPD are binding for Brazilian residents, and you have the specific rights enumerated in Section 12 below.

EEA / UK / Swiss residents (GDPR): If you are located in the European Economic Area, the United Kingdom, or Switzerland, the General Data Protection Regulation (EU) 2016/679 (GDPR) and its national implementations apply to our processing of your personal data. We note that our Article 27 representative has not yet been formally appointed (see Section 1). EEA/UK/CH residents have the rights enumerated in Section 12 below, and may lodge a complaint with the supervisory authority in their country of residence.

US residents (CCPA/CPRA and state laws): If you are a resident of California or another US state with a comprehensive privacy law (Virginia, Colorado, Connecticut, Texas, Utah, Oregon, Montana, Iowa, Tennessee, Indiana, Delaware, New Hampshire, New Jersey, Maryland, or Minnesota), the applicable state privacy statute binds us as a business. The rights specific to US residents are enumerated in Section 12 below.

3. Information We Collect

Cloud services may process the categories below according to the feature used. For local Desktop and Mobile data, also see sections 17 and 18.

We do not intentionally collect special-category personal data (health, biometric, religious, political, or sexual orientation data) through standard platform use. If you include such data in your prompts or knowledge bases, you are responsible for having a lawful basis to do so, and we process it solely to operate the Service as described in this policy.

4. How We Use It

We use personal information for the following purposes:

• Providing and operating the Service: Creating and managing your account, routing your prompts to AI models via OpenRouter, executing tool actions via Composio, storing conversations and knowledge base content, delivering chat responses across all supported channels (hub, public, widget, Telegram, WhatsApp, Instagram, API).
• Billing and payments: Processing subscription payments through Stripe, managing credit balances, sending billing receipts and invoices, handling payment disputes, applying and tracking promotional codes.
• Security and fraud prevention: Detecting and investigating unauthorized access, abuse, and fraudulent activity; enforcing our Acceptable Use Policy; protecting the integrity of the platform and our users; rate-limiting and abuse prevention using visitor IP addresses on anonymous channels (public and widget).
• Product improvement: Analyzing aggregate, anonymized usage metrics to understand how the Service is used and to prioritize new features. We do not use your individual conversation content for product improvement without your explicit consent.
• Transactional communications: Sending account creation confirmations, password reset emails, billing notifications, security alerts, and Service status updates. These are necessary to operate the Service and cannot be opted out of while your account is active.
• Marketing communications: Sending promotional emails, product updates, and feature announcements only if you have given explicit opt-in consent. You may withdraw this consent at any time via the unsubscribe link in any marketing email.

5. Legal Bases (LGPD + GDPR)

For users in Brazil (LGPD) and the EEA/UK/Switzerland (GDPR), every processing activity requires a lawful basis. The table below maps each purpose to its applicable legal basis under both regimes.

Where we rely on consent as the legal basis, you have the right to withdraw that consent at any time without affecting the lawfulness of processing carried out before the withdrawal. Where we rely on legitimate interest, you have the right to object to processing (see Section 12).

6. AI Processing & Subprocessors

In Web agent flows that use OpenRouter, your prompt and conversation context are transmitted over TLS to OpenRouter (OpenRouter Inc., United States), which routes the request to a selected upstream model provider — for example OpenAI, Anthropic, Google, Meta, Mistral, or DeepSeek. The current provider roster is published at /legal/subprocessors.

Your organization chooses which providers may serve available models. The initial setting is All models, which sends data_collection: 'allow': a provider may retain prompts or use them to train models. No data trainingsends data_collection: 'deny' and allows only privacy-verified endpoints with a route accepted by OpenRouter. Zero data retention also sends zdr: true and requires a verified ZDR-compatible endpoint. A model served by several providers remains available when at least one route meets the selected policy. Models without privacy information appear under the initial setting and are unavailable under restrictive settings. Your organization's own OpenRouter account settings may restrict additional models.

In Studio, Meta's meta/muse-spark-1.3-contributor model also requires active user consent in addition to the organization's policy. When allowed, prompts and responses may be used to improve Meta products. Conversation context, history, and tool or file content actually sent to the model may also be part of the request. Allowing collection does not itself specify a retention period: see the model terms on OpenRouter and our subprocessor list. Users can revoke consent in Studio; revocation prevents new requests but cannot undo requests already processed. ZDR concerns model-provider routing; it does not remove SquadOS internal records, operational logs, or data sent through tools and other subprocessors.

Voice-message transcription is an exception. Audio messages are sent directly to OpenAI (United States) for transcription using the Whisper model. Under OpenAI's API data-usage policies, content submitted through the API is not used to train OpenAI's models and is retained only transiently for abuse monitoring before deletion. OpenAI is listed as a subprocessor at /legal/subprocessors.

To power knowledge-base search, text you add to a knowledge base is converted into numerical embeddings (via OpenRouter, model text-embedding-3-small) and stored as vectors in our vector database (Qdrant). When you provide a web address for a knowledge base, its content is fetched on your behalf by Firecrawl.

SquadOS does not train any model on Customer content and does not use your conversation history to fine-tune any model, whether our own or a third party's. AI-generated outputs may be inaccurate, incomplete, or misleading; we do not review the content of individual conversations except to investigate a reported Acceptable Use violation or to respond to lawful legal process.

7. Tool Integration Data (Composio)

SquadOS enables agents to call external tools and services (Gmail, Slack, GitHub, Notion, and many others) through Composio (operated by Composio Inc., United States). When you connect a third-party account, Composio obtains and stores the OAuth token on your behalf. SquadOS stores a reference to that connection, encrypted at rest, only for as long as the connection is active.

When an agent executes a tool action, the action parameters — which may include content from your conversation or knowledge base — pass through Composio to the target third-party service. The results of that action are returned through Composio and stored in your conversation history by SquadOS. Once data reaches the third-party service, that service's own privacy policy governs its use.

If you disconnect a tool integration through the SquadOS dashboard, we will delete the stored connection reference and notify Composio to revoke the associated OAuth token. Token deletion propagates through Composio's systems within 30 days of disconnection. We recommend also revoking the authorization directly within the third-party service to ensure complete removal.

8. Sharing & Disclosure

We share personal information only in the following circumstances:

• Subprocessors: We engage third-party service providers (subprocessors) to help operate the Service. Our complete, current subprocessor list — including the categories of data shared with each — is published at /legal/subprocessors. We provide at least 30 days' prior notice of any new subprocessor addition.
• Legal compliance: We may disclose personal information in response to a valid subpoena, court order, regulatory request, or other lawful legal process. Where legally permitted, we will notify you before disclosing and will challenge overbroad or legally deficient requests.
• Security and fraud investigations: We may share information with law enforcement or other parties as necessary to prevent, detect, or investigate fraud, security incidents, or violations of our Acceptable Use Policy, where disclosure is permitted or required by law.
• Business transfers: In the event of a merger, acquisition, or sale of all or substantially all of our assets, personal information may be transferred to the successor entity. We will provide notice via email and an in-product banner before any such transfer occurs, and we will require the successor to honor this Privacy Policy or give you the opportunity to terminate your account.
• Sale and sharing — CCPA/CPRA: We do not sell your personal information as that term is defined under the California Consumer Privacy Act (CCPA) and its amendment, the California Privacy Rights Act (CPRA). We do not share your personal information for cross-context behavioral advertising purposes. Accordingly, the right to opt out of the sale or sharing of personal information is not applicable, but we acknowledge it here as a stated declaration for the avoidance of doubt.

9. International Transfers

Our primary data processing infrastructure is located in the United States. If you access the Service from outside the United States, your personal information is transferred to and processed in the US. The legal mechanisms we use to legitimize these transfers depend on your jurisdiction.

Brazil (LGPD): For transfers of Brazilian personal data to the United States, we rely on Standard Contractual Clauses (Cláusulas Contratuais Padrão) executed with each subprocessor that handles Brazilian data, as authorized by ANPD Resolution CD/ANPD No. 19/2024. We also rely on the contract-performance basis under LGPD Art. 33 where applicable. Supplementary measures include encryption in transit (TLS 1.2 or higher) and encryption at rest (AES-256 via Supabase).

EEA / UK / Switzerland (GDPR): For transfers of EEA, UK, or Swiss personal data to the United States, we rely on the 2021 Standard Contractual Clauses (Commission Implementing Decision (EU) 2021/914) — Module 2 (controller to processor) or Module 3 (processor to processor) as appropriate — executed with each US-based subprocessor, including Supabase, OpenRouter, OpenAI, Composio, Firecrawl, Qdrant, Meta Platforms, Zoho/ZeptoMail, Stripe, Vercel, and Sentry. Where a subprocessor maintains current certification under the EU-US Data Privacy Framework (DPF), we additionally rely on that certification. As verified on 2026-05-31, Stripe and Vercel are DPF-certified; Supabase, OpenAI, and our other US subprocessors rely on Standard Contractual Clauses. We maintain an internal transfer impact assessment covering our primary subprocessors.

UK transfers are covered by the UK Addendum to the 2021 SCCs. Swiss transfers are covered by Swiss Federal Data Protection and Information Commissioner (FDPIC) guidance on international data transfers. If you have questions about the specific safeguards in place for your jurisdiction, please contact us at [email protected].

10. Retention

We retain personal information for no longer than is necessary for the purposes for which it was collected. The table below describes our retention periods by data category.

After the applicable retention period, data is securely deleted or anonymized. We do not retain personal information in identifiable form beyond the periods listed above except where required by law or ongoing legal proceedings.

11. Security

We implement technical and organizational security measures designed to protect personal information against unauthorized access, disclosure, alteration, or destruction. These measures include: Transport Layer Security (TLS 1.2 or higher) for all data in transit; AES-256 encryption at rest via Supabase; PostgreSQL Row-Level Security (RLS) enforcing per-organization data isolation at the database layer; Role-Based Access Control (RBAC) limiting personnel access to systems and data on a need-to-know basis; optional multi-factor authentication for user accounts; and audit logging of all administrative actions.

We conduct periodic security reviews and work with vendors who perform penetration testing on their platforms. We have an internal incident response procedure that includes containment, eradication, recovery, and post-incident review phases. In the event of a personal data breach, we will notify affected users and relevant authorities in accordance with applicable law: within 72 hours of becoming aware of the breach under GDPR Article 33 where required; within three business days where required under ANPD Resolution No. 15/2024, subject to specific deadlines and legally applicable exceptions; and without undue delay under CCPA/CPRA.

On Web, Sentry session replay is configured to mask all text inputs and displayed text, so conversation content and personal data entered into forms are not captured in session recordings.

No security system is impenetrable. While we take reasonable precautions to protect your information, we cannot guarantee that unauthorized parties will never be able to circumvent our security measures. You are responsible for maintaining the confidentiality of your account credentials and for notifying us immediately of any suspected unauthorized account access at [email protected].

12. Your Rights

Based on where you are located, you have specific rights regarding your personal information. To exercise any of the rights below, contact us at [email protected] or visit /legal/data-deletion.

Brazil — LGPD (Art. 18)

Brazilian residents have the following nine rights under Art. 18 of the LGPD: (i) confirmation of the existence of processing; (ii) access to your data; (iii) correction of incomplete, inaccurate, or outdated data; (iv) anonymization, blocking, or deletion of unnecessary or excessive data or data processed in non-compliance with the LGPD; (v) portability of your data to another service or product provider; (vi) deletion of personal data processed with your consent; (vii) information about the public and private entities with which we have shared your data; (viii) information about the possibility of not providing consent and the consequences of refusing; and (ix) revocation of consent. You may also lodge a complaint with the Autoridade Nacional de Proteção de Dados (ANPD) at gov.br/anpd. We will respond to LGPD requests within 15 days of receipt.

EEA / UK / Switzerland — GDPR (Arts. 15–22)

EEA, UK, and Swiss residents have the following rights under GDPR Articles 15 through 22: (i) right of access (Art. 15); (ii) right to rectification (Art. 16); (iii) right to erasure / right to be forgotten (Art. 17); (iv) right to restriction of processing (Art. 18); (v) right to data portability (Art. 20); (vi) right to object to processing based on legitimate interests or direct marketing (Art. 21); (vii) rights related to automated individual decision-making and profiling (Art. 22). You also have the right to lodge a complaint with the supervisory authority in your country of residence. We will respond to GDPR requests within 1 month of receipt; this period may be extended by a further two months where necessary, in which case we will notify you within the first month with the reasons for the extension.

United States — CCPA/CPRA and State Privacy Laws

US residents in states with comprehensive privacy laws have the following rights: right to know what personal information we collect, use, disclose, and sell (we do not sell); right to delete personal information; right to correct inaccurate personal information; right to opt out of the sale or sharing of personal information (not applicable — we do not sell or share for advertising purposes); and right to limit the use and disclosure of sensitive personal information. We honor the Global Privacy Control (GPC) signal universally — if your browser transmits a GPC signal, we treat it as a request to opt out of any data sharing that requires opt-out under applicable law. We will respond to US state privacy requests within 45 days of receipt; this period may be extended by a further 45 days where reasonably necessary, with notice provided within the initial period.

13. Children

The Service is not directed at individuals under the age of 16. We do not knowingly collect personal information from anyone under 16 years of age. If you are a parent or guardian and become aware that a child under 16 has provided personal information to SquadOS without your consent, please contact us immediately at [email protected].

Upon receiving such notification, we will take prompt steps to delete the child's personal information from our systems and to disable or terminate the associated account. We will not use information collected from a child under 16 for any commercial purpose. If you believe a user of the Service may be under 16, you may report this to [email protected]with the subject line "Minor Account Report."

14. Cookies

SquadOS uses cookies and similar tracking technologies to operate and improve the Service. Cookies are small text files stored in your browser that allow us to recognize your session, remember your preferences, and analyze how the Service is used.

We currently use two categories of cookies and tracking technologies: strictly necessary cookies (required to operate the Service, including the Supabase authentication session and CSRF tokens — no consent required); and aggregate analytics via Vercel Analytics, which collects anonymized page-view and performance data without setting persistent cross-site tracking cookies (legitimate interest). We also use Sentry for error monitoring and masked session replay (legitimate interest); Sentry session replay masks all text inputs and on-screen text so no conversation content is captured. A granular consent banner is forthcoming; until it is deployed, no marketing or targeting cookies are set. For a full cookie inventory and instructions on how to manage or disable cookies, please see our Cookie Policy.

We use PostHog for product analytics and privacy-masked session replay. This helps us understand onboarding, feature adoption, retention, and product friction. We configure PostHog replay to mask all inputs and text and block sensitive product surfaces. Curated product events must not include free-text chat messages, agent prompts, knowledge-base content, documents, tokens, headers, signed URLs, or tool payloads.

15. Changes & Contact

We may update this Privacy Policy from time to time to reflect changes in our practices, technology, legal requirements, or other factors. When we make material changes, we will notify you via email to the address associated with your account and via an in-product banner. The "Last updated" date at the top of this document will always reflect the date of the most recent revision. Continued use of the Service after the effective date constitutes acceptance of the updated policy.

If you have any questions, concerns, or complaints about this Privacy Policy or our privacy practices, please contact us at:

We will acknowledge receipt of your inquiry within two business days and aim to resolve all privacy concerns within 30 days. If you are not satisfied with our response, you have the right to lodge a complaint with the data protection authority in your jurisdiction (see Section 12 for authority contacts by region).

These documents are provided in English and Portuguese. The English-language version is the official and controlling version; in case of any discrepancy or conflict of interpretation, the English version prevails. The Portuguese version is a courtesy translation.

16. SquadOS Web: company collaboration and operations

SquadOS Web is the organization’s collaboration and operations platform. Accounts, members, rooms, messages, agents, knowledge bases, integrations, schedules and usage records are processed in service infrastructure according to the feature used. A Web conversation is not a file stored exclusively in your browser.

Messages and attachments sent to a room are available to authorized participants. Agents may receive history and sources made available to answer; tool results may become part of that history. Knowledge bases may generate indexes and numerical representations of text for retrieval. Publishing an agent, embedding a widget or configuring an external channel creates its own recipients and flows, described in the AI and integration sections.

Meetings may process audio, video and screen sharing to deliver the call. Recording, transcription and other features, when available and activated, involve additional processing. The organization must inform participants of the features used and establish an appropriate legal basis; joining a call does not authorize every subsequent use of its content.

Cookies, product analytics and replay described in sections 11 and 14 refer to web surfaces where those services are configured. Technical interface replay and AI conversation history are different data. Web behavior must not be presumed to apply to Desktop or Mobile.

17. SquadOS Studio Desktop: local files and sessions

Studio Desktop is the assistant that works on your computer, using the Hermes harness integrated into the application. Session history, memory, settings and other local state may persist on the device. Your selected folder is a real folder: tool actions may modify your original files. The application requires an internet connection; local storage does not mean offline operation or no data transmission.

When you request a task, the AI provider may receive your message, instructions, relevant history, memory, attachments, file excerpts, tool results and images actually included in context. Reading a file through the assistant may bring its contents into the next AI request. History compaction, summarization, research, transcription and image generation may require additional calls and different recipients from the main model.

Connecting a SquadOS account does not automatically make every local folder and conversation visible to colleagues. Account features, managed models, research, synchronized memory, shared files, cloud routines and remote access have their own flows. When used, they may transmit content or metadata; they are not covered by a promise that everything stays on your computer.

Provider and connected-service credentials are handled separately from conversation text, using credential storage mechanisms integrated into the application and operating system. This does not make ordinary history, memory or files end-to-end encrypted. Protect access to your device and consider backups, folder synchronization services and system permissions used by you or your company.

Persistent memory allows information to be reused across sessions. Information removed from a conversation may remain in memory, generated files, summaries or external services. Review available memory and synchronization controls and what you include in tasks. An installed skill, extension or tool may execute code or communicate with third parties; its origin and permissions are also part of the privacy decision.

18. SquadOS Mobile: collaboration and remote Studio access

SquadOS Mobile combines organization collaboration and remote access to Studio. Messages, rooms, agents and other collaboration features use the corresponding SquadOS Web cloud services. The Studio area connects to an authorized Desktop session; it must not be confused with an independent assistant running on the phone.

In the remote experience, the computer must be available. Authentication, device discovery, authorization and transport services process the metadata required for the connection. The remote channel carries commands, messages, responses and files used in that interaction. Authentication and protected transport do not, by themselves, guarantee end-to-end encryption against every intermediary.

The phone may store a login session, preferences, drafts, prepared attachments and temporary previews in application storage. Drafts may persist for recovery; locking the screen or leaving a conversation does not mean deleting them. Sending an attachment transfers the selected content into the task flow, and saving it to the gallery or sharing it creates a copy outside private application storage.

Camera, microphone, photos, files and notifications depend on the features used and system permissions. You can review these permissions in phone settings. Notification registration may associate a delivery token with the account or device and involve platform delivery services. Notifications may appear on the lock screen: configure system previews according to the sensitivity of your work.

Mobile technical reports use their own filters to reduce personal data and free-form content. The Desktop telemetry control does not automatically disable Mobile reports, Web analytics or notifications. Features described in this policy depend on availability in your installed version; this description does not announce a feature as released.

19. Organizations, permissions and sharing

CSC Labs LLC determines the purposes and means of processing necessary for its direct relationship with users, including accounts, billing, security and product operations. When an organization determines how employee, customer or contact data is used in the service, it may act as controller and SquadOS as processor for that processing, according to the feature and contract. One relationship may contain both situations; the role is not determined solely by the plan name.

Administrators may manage members, policies, models and administrative information according to their permissions. This does not mean unrestricted access to all personal Desktop conversations. Room, agent, knowledge-base and shared-file content follows the permissions of the resource. Monitoring policies or tools the company installs directly on the computer are the company’s responsibility and are not neutralized by Studio.

Sharing a Studio folder is a specific action, separate from connecting an account. Organization sharing features may upload files and versions to private cloud storage along with relative names, sizes, hashes, version identifiers and access records. The file architecture uses Cloudflare R2 for objects and SquadOS/Supabase services for authorization and metadata. Permissions may grant viewing or editing by organization, team or person; check the effective audience before sharing.

Private storage and time-limited access links reduce public exposure but do not make a file inaccessible to systems authorized to process it. Revoking permission prevents new authorized access according to the feature; it does not erase copies a recipient has already downloaded, exported, photographed or forwarded.

Changing organizations, removing an employee or closing an account may have different consequences for shared company data and personal local data. The organization must establish its retention, offboarding and work-continuity rules. Requests concerning data controlled by the organization will be forwarded or handled in cooperation with it, without excluding statutory rights.

20. AI providers, training and data routes

SquadOS’s commitment not to train models on Customer content concerns our use of that content. It does not replace the terms of the AI provider, router or integration. On Desktop, personal OpenAI or Claude connections follow the product, plan and settings of that account. SquadOS-managed models follow the route and policy applicable to the organization.

We do not presume that a personal account has the same conditions as an API or business account. Training, retention, safety review and processing locations may vary. Organization model policy may restrict managed routes, but must not be interpreted as universal control over personal accounts, web research, connectors, transcription or external tools.

No training means a restriction on content use for training on the covered route. Zero data retention means a provider retention condition on the covered route, subject to applicable conditions and exceptions. Neither option automatically deletes SquadOS history, local files, memory, backups, billing records or data sent by tools.

Before sending sensitive data, identify the model, account and integrations used. A tool result may return to AI context; research may transmit terms from your request to a search service. Changing the main model does not necessarily change providers for every auxiliary step. You can consult service providers in the subprocessor list and the terms of services you connect directly.

21. Desktop telemetry and technical reports

Optional Studio Desktop diagnostics are enabled by default and can be disabled in application settings. Being enabled by default must not be understood as express consent for any purpose. Diagnostics are intended to identify failures, assess performance and improve product operation; they do not authorize using conversations or documents for model training.

Usage counters accept only categories and values defined in a closed contract. Durations, counts and memory usage are reduced to ranges. The contract does not admit messages, file contents, person or project names, paths, commands, URLs or account identity. Events that do not meet the contract are discarded. The counter queue stays in memory and is discarded after a failed send, shutdown or disabling.

Error reports are a separate flow, sent to Sentry when configured and allowed. They may contain application version, environment, technical failure types and codes, and filtered execution frames. Sending is conditioned on the current setting. Desktop configuration does not attach screenshots or breadcrumbs, and the filter removes event attachments, including memory dumps, before admitting the report.

The error queue differs from the counter queue: a local cache of sanitized events may exist for later delivery while the option is enabled. Disabling prevents new optional sends; it does not retract events already received by services. Local technical records and system crash artifacts may exist even without remote sending and should be considered when cleaning the device.

Network transport may expose IP and connection metadata to the receiving service and delivery infrastructure even when the event body does not identify a person. Event minimization therefore does not mean an absolute absence of personal data throughout the chain. Encryption protects data in transit and in storage where applied; a recipient that needs to process it may access the content.

Disabling optional diagnostics does not prevent AI calls, authentication, synchronization, billing, notifications or records necessary for security and operation of the features used. These flows have separate purposes. The Desktop control does not change preferences or records of other products and providers.

22. Computer control, browser and tools

When you authorize a control tool, it may observe the necessary interface or content and take actions on the computer or browser. Captures used for the task may enter model context. This differs from a telemetry screenshot: excluding screenshots from error reports does not prevent images being used in an authorized control task.

An authenticated tab may contain customer data, email, documents or internal systems. A task may transmit information to involved services or perform an action with external consequences. Authorizing control of a surface must not be interpreted as unrestricted access by colleagues or support; recipients remain those of the applicable flow and permissions.

Local tools run within the effective limits of the system and integration. A working folder, an instruction written to the model or visual separation between sessions is not an operating-system security boundary. Review authorizations, tabs and files used, and stop control when unnecessary. We do not describe local execution as isolated solely because a selected folder exists.

23. Retention and deletion by storage location

The table in section 10 concerns the cloud-service records listed there. It does not mean that history, memory and files on your device will be erased within the same period. Local data follows application controls, actions taken on the device and backups configured by you or your organization.

Deleting a conversation is not the same as deleting a file the assistant created, extracted memory, a message sent to third parties or a shared copy. Disconnecting an account stops that connection but does not imply deleting all local history. Uninstalling an application may leave data, credentials, exports and backups: check the corresponding locations and controls.

Shared files may have versions and trash; cloud data may remain in backups for their recovery window or be retained for legal obligations. Data at an AI provider or connected service also follows that provider’s deletion rules and mechanisms. Deletion in SquadOS is not a promise of simultaneous erasure at every third party.

To request deletion, identify the product, account, organization and data category concerned without sending passwords or credentials. We will explain the scope of the action, necessary identity verification and applicable exceptions. We do not present access revocation, stopping training or removal from a screen as proof that every copy has been erased.

24. Your choices and transparency

You can disable optional Desktop diagnostics, review the model and account used, manage integrations, check memory and sharing, stop remote control and adjust phone permissions and notification previews according to available controls. The organization can configure features and policies it administers. These choices are independent: disabling one feature does not automatically reset the others.

If you are unsure who receives data or who can access it, request clarification before including confidential information. Do not send credentials, full documents or customer data to support when a description of the failure is sufficient. Content you choose to send to support becomes part of that interaction; we do not treat it as general authorization to read your local sessions.

The measures described in this policy are not a security certification, a guarantee against incidents or a general promise of end-to-end encryption. Changes to purposes, recipients and controls must be reflected in this policy and communicated according to applicable obligations. Your section 12 rights remain valid regardless of which product you use.