privacy policy

1. Introduction

Rabbit Inc. (“Rabbit,” “we,” “us,” or “our”) respects your privacy and is committed to protecting it through our compliance with this Privacy Policy (this “Policy”).

This Policy describes the types of information we may collect from you or that you may provide when you use rabbit OS3 (“OS3”), the rabbit agent application, the rabbit r1 (“r1”) and other rabbit-branded devices, our websites, and the related software and services we provide (collectively, the “Service”). It also explains how we use, share, retain, transfer, and protect that information, and the rights and choices available to you.

Data Controller. Rabbit Inc., a Delaware corporation based in the United States, is the data controller responsible for the processing of your personal information under this Policy. Our contact details, including our Data Protection Officer and EU/UK representative, are set out in Section 14.

This Policy applies to information we collect:

• On or through OS3 and our websites (including rabbit.tech), the rabbit agent application installed on your computer, and any other software or interface we provide;

• Through paired Channels, including Telegram, Messages (by iMessage, RCS, and SMS), and other channels we may add;

• Through rabbit-branded hardware, including the r1 device and future devices, when paired with your account;

• In email, text, and other electronic messages between you and us; and

• When you register for an account, pair a Node or Channel, configure a model, or otherwise use the Service.

It does not apply to information collected by:

1. third-party models, applications, websites, tools, or services that you configure or invoke yourself, including BYOK (Bring Your Own Key) Models and third-party Skills, which are governed by their own privacy policies;

2. any website or service not operated by Rabbit, even if linked from the Service; or

3. your employer or organization where you use the Service through an enterprise account administered by that organization.

2. Notice and Children’s Privacy

Please read this Policy carefully to understand our policies and practices regarding your information and how we will treat it. If you do not agree with our policies and practices, your choice is not to use the Service. By creating an account, installing the rabbit agent, pairing a device or Channel, clicking “I agree,” or otherwise accessing or using the Service, you acknowledge that you have read, understood, and agree to this Policy.

Notice to California Residents. If you are a California resident, this Policy provides the notices required under the California Consumer Privacy Act (“CCPA”), as amended by the California Privacy Rights Act (“CPRA”), including the categories of personal information we collect, the sources and purposes of collection, the categories of recipients, your rights to know, delete, correct, limit the use of sensitive personal information, and opt out of sale or sharing, and the right to non-discrimination. See Section 9 (California Privacy Rights) for details.

Not for Use by Children. The Service is not directed to children. We do not knowingly collect personal information from anyone under the age of 18. The Service is intended for users aged 18 and over. If you are under 18, do not use the Service or provide any personal information to us.

If we learn we have collected personal information from anyone under the age of 18, we will delete that information promptly. If you believe we might have information from or about a child, please contact us using the details in Section 14.

3. Information We Collect About You

“Personal data” or “personal information” means any information relating to an identified or identifiable individual, as defined under the General Data Protection Regulation (“GDPR”), the UK GDPR, and applicable U.S. federal and state privacy laws. It does not include data where the identity has been removed (anonymous data).

We collect, use, store, and transfer different kinds of personal data about you, which we have grouped together as follows. Where a field is listed, the purpose for which it is collected is stated alongside it.

3.1 Account and Registration Information

When you register and log in to your Rabbit account, we collect the following information, which you provide directly to us:

• Email address: used for account registration, login authentication, password recovery, transactional notifications, and Service communications;

• Account identifier (user ID / subject ID): used to identify your account across the Service, pair Nodes and Channels, and associate your conversation history and Memory with your account;

• Account password: managed by our identity provider (Auth0/Okta); passwords are not stored by Rabbit in plain text and are used only for authentication;

• Onboarding and profile information: where you provide it, your display name, reported city or region, and time zone, used to personalize your experience and configure regional settings;

• Device display names and labels: names you choose for paired Nodes and Channels, used to help you identify and manage your devices.

3.2 Node Pairing and Device Environment Information

When you install the rabbit agent on a computer and pair it with your account (thereby making that computer a “Node”), we collect the following information:

• Node identifier and public key: a unique identifier and a locally generated public key for the Node, used to authenticate, route instructions to, and verify the identity of the paired device; the private key is generated and retained on your device and is not sent to us;

• Connection and heartbeat status: used to determine whether the Node is online and available to receive instructions;

• Operating system and distribution, architecture, shell, hostname, OS username, runtimes, executable path, CPU cores, memory, locale, and virtual-machine status: reported by the rabbit agent, used to select compatible tools and commands, route tasks to the correct device, and diagnose environment-specific issues;

• rabbit agent version: used to determine whether updates are needed and to manage feature availability;

• Working directory and default file paths: where you configure them, used to locate files you ask OS3 to access;

• Learned device notes (OS3 Note): in addition to labels you choose, OS3 may maintain notes about each paired Node learned from task context, used to select the correct computer and reuse device-specific knowledge.

3.3 Conversation, Input and Task Data

When you interact with OS3 through a Channel, we collect the content of your communications and the tasks you request:

• Chat input and instructions: the contents of your conversations and instructions sent through the website, Telegram, Messages, or a paired r1, including text, pasted content, and messages, used to understand your intent, carry out your tasks, and maintain conversation history; unrestricted text input may contain personal information you choose to include;

• Uploaded files and attachments: files you select, drag, paste, or send from a Channel or r1, including file bytes and filenames, used for processing in connection with a task you request;

• Task records and context: records of the tasks you initiate, the prompts and context assembled for the model, tool definitions and tool-call results. The AI model you use (your BYOK Model or a model provided by Rabbit) generates the Output, which is then returned to you and used to execute tasks, provide continuity, debug, and improve the Service.

3.4 Device-Control and Desktop Operation Data (DLAM)

When a task requires OS3 to operate your computer through the device-control capability (“DLAM”), the following information may be generated or collected:

• Tool arguments and results: the arguments passed to local tools or commands and the results returned, used to execute and verify the task;

• Command output (stdout/stderr): output from terminal commands or scripts executed on your Node at your direction, used to complete the task and diagnose errors;

• Screenshots and dynamic on-screen content: when a task uses the visual fallback (i.e., it needs information displayed on screen that is not otherwise available), the rabbit agent captures screen content. This is processed transiently and is not permanently retained by default;

• Input simulation records: records of mouse movements, clicks, and keyboard input simulated by the rabbit agent, used to execute the requested operation and for audit and debugging.

3.5 File and Content Data Accessed for Tasks

When a task requires access to files on your computer, the rabbit agent locates and accesses only the specific files and data the LLM determines are needed for that task within your instructions and authorized scope. The content of those files is processed through Rabbit’s servers together with the prompts needed for the task and is sent to the LLM provider you have connected, where it is handled under that provider’s terms. Rabbit does not keep a copy of the file itself; excerpts or results derived from it may be retained in your conversation history, Memory, or operational logs as described in this Privacy Policy. The original file remains on your computer.

3.6 Memory and Inferred Profile Data

OS3 may build Memory from your conversations to provide continuity and personalization across tasks and Channels. We collect:

• Memory entries: facts, persons, entities, keywords, locations, topics, and other information of reference value extracted from your conversations, used to maintain continuity across sessions;

• Embedding vectors: vector representations used to recall relevant Memory entries when you start a new task;

• Source conversation references: references to the conversations from which Memory entries were derived, used for transparency and to help OS3 locate Memory entries when you ask it in a conversation to correct or forget them;

• SOUL/persona document: in addition to Memory entries, we store a SOUL document containing profile and personality text that you provide (such as a preferred user name or butler name) and information derived by AI from your conversations and onboarding, used to personalize replies and retain stable user preferences.

3.7 BYOK Model Credentials and Channel Pairing Information

When you configure a BYOK Model or pair a Channel, we collect:

• BYOK Model provider name, endpoint, and API key: used to route your Input to the model provider you designate. API keys are encrypted using envelope encryption with AWS Key Management Service, and only a masked version is displayed in the interface;

• Phone call credentials: the OpenAI API key and project ID you connect in the Phone calls section of OS3 settings, used to process your phone calls through your own OpenAI account. These credentials are separate from your BYOK Model API key and are encrypted in the same way;

• Telegram pairing information: if you pair Telegram, your Telegram user/bot identifiers, bot ID and bot token needed to link that Channel, plus messages and files you send through that Channel. The bot token is stored with your account (protected by our database-level encryption) and remains on our servers until you unpair. Unpairing removes the OS3 pairing but does not delete messages you have exchanged with the bot on Telegram itself;

• Messages information: if you use Messages as a Channel, we collect the identifiers used to link the Channel (phone number and/or Apple ID email), the messages and attachments you send and receive through Messages, and delivery status. Messages are relayed between you and OS3 by our messaging provider (Photon). Messages pairing information remains on our servers until you unpair the Channel. Unpairing removes the OS3 pairing but does not delete messages you have exchanged through Messages on your devices or with our messaging provider (retained per its retention policy);

• Phone call information: when you use OS3 to place a phone call, we collect the phone number you instruct OS3 to call, call metadata (such as time, duration, and connection status) and a transcript of the call generated by OpenAI. We do not record or store the raw audio of calls on Rabbit’s servers. The content of a call exists as a text transcript generated by OpenAI and included in your conversation history. The transcript may include the speech of both you and the called party. Where required by applicable law, OS3 will disclose at the beginning of each call that the call is being conducted by an AI assistant and may be recorded or transcribed. Phone calls are processed by OpenAI under your own OpenAI account, using the OpenAI API key and project ID you connect in the Phone calls section of OS3 settings. This is separate from the BYOK Model you use for general OS3 tasks. OpenAI processes the call audio in real time and generates the call transcript; this usage is billed to your OpenAI account. Twilio and Photon carry call audio in transit for call delivery; whether either provider records or retains call audio is subject to that provider’s privacy policy.

3.8 Log, Usage and Diagnostic Data

We automatically collect the following information when you use the Service:

• IP address, browser type and version, operating system, and device identifiers: used for security, fraud prevention, rate limiting, and to ensure compatibility;

• Access times and usage/token-consumption counters: used for rate limiting, Service planning, and billing where applicable;

• Operational and diagnostic logs: correlation/trace identifiers, request and error metadata, timing, and, during the technical preview, debug log content that may include chat, tool, and prompt text (retained for a limited period for troubleshooting);

• Node lifecycle audit logs: records of Node registration, pairing, deregistration, and permission changes, used for security audit and traceability;

• Normalized email-derived bucket hash: a hash derived from your email address, used together with your IP address for rate limiting and abuse prevention (for example, to deter verification-email abuse, brute-force and password-guessing attempts), retained for approximately 15 minutes;

• Onboarding progress and diagnostic data: the onboarding step you have reached, attempt counts and timestamps, and error codes, associated with your user ID, used to track and recover the onboarding flow and diagnose drop-off points;

• HTTP tunnel and local application proxy: where you use the browser-based Node preview or a local model endpoint, proxied HTTP/WebSocket traffic to your Node is relayed through our edge; this traffic is streamed transparently and is not persistently stored, and access is authorized by sealed short-lived tokens;

• Coding-agent and Skill import: where you opt in through an import card, OS3 detects installed coding agents and reads selected reusable instructions, commands, project instructions, fixed preferences and source paths (excluding credential files and transcript contents), converting them into OS3 Skills or profile facts.

3.9 Rabbit Hardware (r1) Data

When you pair an r1 or other rabbit-branded hardware with OS3, we receive:

• Device identifiers: serial number, hardware model, firmware version, and connection status, used for device identification, pairing, and management;

• Content you upload from the device: photos, voice, messages, or other content you choose to transmit from the r1 to OS3 for processing;

• Device usage and diagnostic data: usage statistics, error logs, and performance information, used to diagnose faults and improve the device;

• r1 data export: where you use “export data” in OS3 settings and select “only r1 data”, we export all your r1 data except conversations — including Journal entries, Hoover or Intern sessions, and referenced photos, audio, images, documents or artifact files (earlier rabbithole data and newer r1 data are included in the same download) — as a ZIP to a temporary AWS S3 bucket for download. The download link is available for 7 days, and the signed URL expires after 1 hour.

3.10 Information from Third Parties

We receive information from third parties in the following circumstances:

• Identity provider: we receive authentication information, account identifiers, and email verification status from our identity provider (currently Auth0/Okta);

• LLM and embedding providers: where you use a BYOK Model or the embedding service, the provider may return usage metadata. For BYOK Models, the provider processes your Input under its own terms;

• Telegram: when you pair Telegram as a Channel, we receive the messages and files you send through that Channel;

• Messaging provider (Photon): if you use Messages as a Channel, we receive from this provider the messages, attachments, and delivery status associated with your Messages Channel;

• Where applicable, payment processors and hardware fulfillment providers: we receive order and shipping information necessary to process purchases of rabbit hardware or paid services; for paid services, our payment processor Stripe provides us with payment status, transaction and subscription details, and limited billing information; we do not receive or store your full payment card number;

• For phone calls placed through OS3: we also collect the called party’s phone number (an identifier) and call transcripts (audio/electronic information). This information may be shared with Twilio and Photon for call delivery and with OpenAI, under your own OpenAI account, for real-time call audio processing and transcription, as described in Section 7.

3.11 Aggregated Data and Optional Workspace Scan

Aggregated Data. We also collect, use, and share aggregated data, such as statistical or demographic data, for any purpose. Aggregated data may be derived from your personal data but is not considered personal data under applicable law, as it will not directly or indirectly reveal your identity. However, if we combine or connect aggregated data with your personal data so that it can directly or indirectly identify you, we treat the combined data as personal data and will use it in accordance with this Policy.

Optional Workspace Scan. Where you opt in during setup, the rabbit agent may inspect recent file names, paths, and modification metadata, and the list of installed applications on your Node, in order to infer work context and initialize a work-area profile. Scan results are processed in the same way as other tasks you give OS3: they are sent to Rabbit’s servers for processing and may become part of your Memory. They are not stored separately, and your files themselves are not stored on Rabbit’s servers.

3.12 Sensitive and Special-Category Information

You may choose to include in your Input, or OS3 may need to access, sensitive personal information (for example, health, financial, precise location, biometric data, or government identifiers) or GDPR Article 9 special categories of personal data in the course of a task. We do not seek to collect such information, but because you may include it in unrestricted text input or it may appear in files or on screen during a task, it may be processed incidentally. Where we process special categories of data, we rely on your explicit consent or another applicable legal basis. You are responsible for not including sensitive information you do not wish to be processed, and for closing or hiding sensitive windows or documents before initiating a task that uses the visual fallback.

3.13 Cookies and Similar Technologies

We use first-party cookies, browser/local storage, and (only on www.rabbit.tech and only where you opt in) a third-party advertising pixel. Where applicable, our payment processor Stripe also sets strictly necessary payment and fraud-prevention cookies when checkout or payment pages are served on our Sites (as defined in our Cookie Policy). Strictly necessary cookies are used to authenticate and secure your session, maintain login state, and enable core functionality. The principal cookies and similar technologies are summarized in our Cookie Policy.

4. How We Use Your Information

We use information that we collect about you or that you provide to us, including any personal information, in the following circumstances:

• To provide the Service, including receiving your instructions through Channels, coordinating agents, Skills, models, and devices, operating your Node, accessing the specific files and tools needed to complete your tasks, and returning results to you;

• To enable account registration, login authentication, Node and Channel pairing, and management of your account;

• To maintain conversation history and Memory so that OS3 provides continuity across Channels and devices;

• To operate the device-control (DLAM) capability, including simulating input, accessing files and applications, and executing commands at your direction;

• To route your Input and necessary context to the BYOK Model you configure, and to return Output to you;

• To provide customer and technical support, verify and deliver hardware orders, and respond to your inquiries;

• To maintain security, prevent fraud and abuse, enforce rate limits, and audit Node lifecycle actions;

• To diagnose problems, maintain reliability, and analyze performance and usage;

• To fulfill any other purpose for which you provide the information;

• To carry out our obligations and enforce our rights arising from any contracts entered into between you and us, including for billing and collection;

• To notify you about changes to the Service or this Privacy Policy through in-Service notification or other means described in Section 13;

• To review, improve, and develop the Service, including improving and training our models where you have not opted out (see Section 6.4);

• To comply with legal obligations, respond to lawful requests, and protect rights, property, and safety; and

• For any other purpose with your consent.

5. Our Legal Basis for Processing Your Information

Where the GDPR or UK GDPR applies, we may only use your information when we have a “legal basis” to do so under Article 6 (and, where relevant, Article 9). We use different legal bases depending on the purpose of the processing. These legal bases are described below. For more information on how to exercise your rights, please see Section 8 (Your Rights and Choices).

5.1 Contractual Necessity

We use your Account Information, Node pairing information, conversation and task data, and file content accessed for tasks where it is necessary to perform the contract you enter into (for example, our Terms of Use) when you register for, access, or use the Service. This includes processing to:

• Provide you with the Service, including logging in, pairing and controlling Nodes, sending instructions, receiving Output, and using OS3 across Channels;

• Enforce our Terms, including suspending or restricting access if we determine that you are violating our Terms;

• Administer the Service, including communicating with you on Service-related matters and responding to your queries.

Your rights: Whenever we use your information on the basis that it is necessary for the contract we have with you, you have the right to port information you have provided to us (see Section 8.4).

5.2 Legitimate Interests

We use your information where this is necessary to achieve legitimate interests, whether belonging to us, you, or a third party, provided these interests are not outweighed by your interests or fundamental rights and freedoms. Where we rely on legitimate interests, we conduct a balancing assessment. The legitimate interests we rely on include:

• Ensuring the security and stability of the Service, including identifying and combating technical or security issues, detecting abuse, fraud, and illegal activity, and enforcing rate limits;

• Improving and developing the Service, including analyzing usage patterns, diagnosing faults, improving navigation and task-execution algorithms, and improving and training our models where you have not opted out (see Section 6.4);

• Administering and protecting our business, including enforcing our Terms, managing corporate transactions, and protecting our rights, property, and safety;

• Providing effective customer support and troubleshooting.

Your rights: Whenever we use your information based on legitimate interests, you can object to, and request restriction on, such usage (see Sections 8.5 and 8.6).

5.3 Your Consent

We use your information based on your consent where we are not relying on another legal basis, including:

• Optional advertising cookies on www.rabbit.tech (you may refuse them in the consent banner and change your choice as described in our Cookie Policy);

• Processing of special categories of personal data (GDPR Article 9) where you include such information in your Input or it is accessed during a task;

• Marketing communications, where you have opted in;

• Optional workspace scan or coding-agent import features, where you authorize them.

Your rights: Whenever we use your information based on your consent, you can withdraw your consent at any time. However, your withdrawal of consent will not affect the lawfulness of processing your information based on your consent before its withdrawal (see Section 8.7).

5.4 Compliance with a Legal Obligation

We may use your information where it is necessary to comply with a legal obligation. This includes:

• Complying with applicable laws, regulations, court orders, and lawful government or regulatory requests;

• Maintaining records required by law, including tax, accounting, and audit records;

• Complying with valid legal requests such as orders from law enforcement agencies or courts.

We generally use account records, logs, and transaction information for these purposes, although the information used depends on the specific situation.

5.5 Protection of Vital Interests

We may use your information where it is necessary to protect your or someone else’s life, physical integrity, or safety. This includes providing law enforcement agencies or emergency services with information in urgent situations.

6. AI Models, Memory and Automated Processing

6.1 BYOK Models

When you configure a BYOK Model with your own API key, we store your API key encrypted on our servers (see Section 3.7) and use it solely to authenticate and route your Input to the model provider you designate. Your Input is routed through Rabbit’s servers to your selected BYOK Model provider, and the Output is returned to Rabbit’s servers and delivered to you. We process request payloads transiently to route them and do not persistently store them, except that your conversation history (including Input and Output) is stored as described in Section 3.3 and limited operational logs are kept as described in Section 3.8. The model provider processes your Input under its own terms. To index Memory entries so that OS3 can find relevant entries more quickly and accurately, we generate embeddings through an embedding provider (currently OpenAI text-embedding-3-small). The embedding provider is used only for indexing; it is not the main mechanism by which OS3 remembers information.

6.2 What Is Sent to Model Providers

To complete a task, OS3 sends to the selected LLM provider your prompts, conversation history and context, the content of selected files or images decoded into the request, tool definitions and tool-call results. The model generates the Output, which is then returned to you. These request payloads are transmitted to the model provider over TLS and are not persistently stored on Rabbit’s servers; only the resulting conversation turns are persisted in our database as described in Section 3.3. Where you request image generation, your prompt and generation settings (model, size, quality) are sent to the selected image-generation provider.

Where Rabbit selects the model for you — for your subscription allowance, usage credits, and free trial credits — the AI model provider that serves the request processes your Input and Output under its own terms, which may include using them to train its models. That provider may change from time to time, and is reached directly or through an AI gateway.

When OS3 places a call, the call audio is streamed in real time to OpenAI under your own OpenAI account (separate from any BYOK Model you configure), which generates the transcript and speech on the call. OpenAI’s processing is governed by OpenAI’s own terms and privacy policy, and this usage is billed to your OpenAI account.

6.3 Memory Feature and Your Controls

OS3 provides cross-session Memory so that your experience is coherent and personalized. Based on your conversations, OS3 automatically extracts and stores information of reference value (such as your preferences, projects, frequently used files, and contacts). Memory is part of how OS3 works. It cannot be turned off; individual Memory entries cannot be deleted separately, and OS3 settings do not include separate controls for Memory.

You can manage your Memory in two ways:

1. You can manage Memory by telling OS3 in a conversation — for example, asking it to remember something, to correct something it remembered incorrectly, or to forget something. These requests are carried out by the AI model you use (your BYOK Model or a model provided by Rabbit), so they may not always succeed or may be misunderstood. Clearing your conversation in OS3 settings does not delete Memory.

2. You can delete Memory together with all your other data or your account. Memory, including its embeddings, is permanently deleted when you choose either “delete all your data” or “delete account” in the Account section of OS3 settings. The “delete all your data” option erases all data stored under your account while keeping your account, after which you start setup again; the “delete account” option does the same and also closes your account itself, including your sign-up email. Neither action can be undone. See Section 12.

6.4 Use of Content to Improve or Train Models; Your Opt-Out

We are careful with your conversations and files. We do not sell your conversation or file content. Where we wish to use Input or Output to train, fine-tune, or improve our models, we will do so only where you have not opted out and where permitted by applicable law. You can opt out of the use of your conversation content for model improvement in OS3 settings (in the OS3 settings panel, under the “account - model training” section, select to turn off “help improve rabbit’s models”). Opting out will not affect your ability to use the Service.

That setting covers our own use of your content only. It does not control what an AI model provider selected by Rabbit does with the content it receives, which is governed by that provider’s own terms as described in Sections 6.2 and 7.1. Some of those providers offer no way to opt out of their own model training, and where a provider offers no such choice, your setting in OS3 cannot change what that provider does. If you want to make that choice yourself, use a model you configure yourself — a BYOK Model, another provider you configure, or a local model — which is processed under the terms you have with that provider (see Sections 6.1 and 7.2).

6.5 Automated Decision-Making and Human Oversight

OS3 automates tasks on your devices at your direction. It is not designed to make, and we do not use it to make, decisions that produce legal or similarly significant effects about you (such as credit scoring, employment decisions, or access to essential services) based solely on automated processing. In accordance with applicable transparency requirements, you are informed that you are interacting with an AI system and that Output is generated by an AI model. You may request human review of any automated outcome by contacting us using the details in Section 14.

6.6 Nature and Ownership of Output

LLM Output may be incorrect, incomplete, or unsuitable, and may resemble existing content. You are responsible for reviewing Output before acting on it and for the decisions you make based on it. As between you and Rabbit, and subject to the applicable model-provider terms, rights in your Input and in Output generated using a BYOK Model belong to you.

7. How We Share and Disclose Your Information

We may disclose aggregated information about our users, and information that does not identify any individual, without restriction. We may disclose personal information that we collect, or you provide as described in this Policy:

7.1 Service Providers and Cloud Infrastructure

We use carefully selected recipients who process on our behalf under written agreements or as independent providers. The principal recipients are:

7.2 Your Own Model and Tool Providers

When you use a BYOK Model, a Skill, a connector, or a third-party application, information is sent to that provider through Rabbit’s servers as needed for the task, under its own terms and privacy policy. Rabbit is not responsible for the processing practices of those providers, and you should review their policies before configuring them, except where a local Skill is configured to communicate directly.

7.3 Corporate Group and Affiliates

We may share your information with our corporate affiliates and subsidiaries where necessary to provide the Service, subject to appropriate confidentiality and data-processing agreements.

7.4 Corporate Transactions

Your information may be disclosed to third parties in connection with a corporate transaction, such as a merger, sale of assets or shares, reorganization, financing, change of control, or acquisition of all or a portion of our business. In such a case, your information may be transferred as part of the transaction, subject to confidentiality and to this Policy.

7.5 Legal Obligations and Rights

We may access, preserve, and share the information described in Section 3 with law enforcement agencies, public authorities, or other third parties if we have a good faith belief that it is necessary to:

• Comply with any court order, law, or legal process, including responding to any government or regulatory request, as consistent with internationally recognized standards;

• Enforce or apply our Terms of Use and other agreements, including for billing and collection purposes;

• Protect the rights, property, or safety of Rabbit, our customers, or others, including protecting life or preventing imminent bodily harm;

• Investigate potential violations of and enforce our Terms, or detect, investigate, prevent, or address misleading activity, copyright infringement, or other unlawful activity.

7.6 With Your Consent

In addition to the purposes described in this Policy, we may disclose your personal information for any other purpose for which you have given us your explicit consent. You may withdraw your consent at any time as described in Section 8.7; such withdrawal does not affect the lawfulness of processing based on consent before its withdrawal.

8. Your Rights and Choices

You have rights and choices when it comes to your information. Some of these rights apply generally, while others will only apply in certain circumstances. Depending on the scenario, these rights may be subject to some limitations and exceptions under applicable law. For users in the EEA, UK, and Switzerland, the rights below apply under the GDPR / UK GDPR. For users in the United States, additional rights under the CCPA and other state laws are described in Section 9.

8.1 Access Your Information

You may access the personal information we process about you directly in OS3, where you can view your account information, your paired Nodes and Channels in OS3 settings, and view your current and past conversations in OS3. You may also ask OS3 in a conversation what it remembers about you.

8.2 Delete Your Information

You can delete your information as described below. In the Account section of OS3 settings, choose “delete all your data” to permanently erase all data stored under your account while keeping your account, after which you start setup again; or choose “delete account” to do the same and close your account itself, including your sign-up email (see Section 12).

You can also unpair Nodes and Channels and remove BYOK API keys in OS3 settings, and you can ask OS3 in a conversation to forget specific information, though such requests may not always be processed reliably; individual Memory entries cannot currently be deleted separately.

Clearing your conversation in OS3 settings removes it from your view only; it does not delete your conversation history or Memory from our servers. Deletion requests are subject to identity verification and to the limited exceptions described in this Policy (for example, information we are required to retain by law).

8.3 Rectify Your Information

You may also change your account information where that information is not accurate or complete in OS3 settings. You can ask OS3 in a conversation to correct information it has remembered about you (see Section 6.3); such corrections are carried out by the AI model and may not always succeed. If you cannot correct inaccurate information yourself in OS3, please contact us using the details in Section 14.

8.4 Port Your Information

You have the right to data portability in circumstances where we rely on contractual necessity and consent as our legal basis. This means that you have the right to receive your information in a structured, commonly used, machine-readable format and to transmit it to another controller.

You may exercise your right to access and data portability by using “export data” in the Account section of OS3 settings. Select “all data” to export your OS3 data (including conversation content, Memory, and associated embeddings) and your r1 data (including intern sessions, journal, photos, audio and artifacts, earlier rabbithole data and newer r1 data), or select “only r1 data” to export your r1 data only.

Alternatively, you may ask OS3 in a conversation to export your conversation history and Memory; exports generated in this manner are produced by the AI model and may not be complete or fully accurate.

8.5 Object to the Processing of Your Information

You have the right to object to the processing of your information in certain circumstances. The right to object applies when the processing you are objecting to is based on legitimate interests or is for direct marketing. In submitting an objection request, you should explain the basis for your objection so that we can assess whether our interests in processing override your rights and freedoms. To object to the use of your conversation content for model improvement, you can use “help improve rabbit’s models” in OS3 settings.

8.6 Restrict the Processing of Your Information

You have the right to request the restriction of the processing of your information where:

1. You are challenging the accuracy of the information;

2. The information has been unlawfully processed, but you are opposing erasure;

3. We no longer need the information for the purposes of processing, but you require it for the establishment, exercise, or defense of legal claims; or

4. You have objected to processing based on legitimate interests, pending verification of whether our grounds override your objection.

8.7 Withdraw Consent

Where we rely on your consent to process your personal information, you have the right to withdraw your consent at any time. You can withdraw your consent to the use of your conversation content for model improvement yourself at any time in OS3 settings (“help improve rabbit’s models”). If you cannot withdraw a consent yourself in OS3, please contact us using the details in Section 14. Withdrawal of consent will not affect the lawfulness of processing based on consent before its withdrawal.

8.8 Automated Decisions

Under GDPR Article 22, you have the right not to be subject to a decision based solely on automated processing, including profiling, that produces legal effects concerning you or similarly significantly affects you. OS3 operates in response to your instructions and does not make automated decisions about you that produce legal or similarly significant effects, as further described in Section 6.5. If you believe an automated outcome has produced such an effect, you may request human intervention, express your point of view, and contest the decision by contacting us via the email address described in Section 14.

8.9 Exercising Your Rights

Many of the above choices are available to you directly in OS3 (see Sections 8.2, 8.4 and 8.7). For any right you cannot exercise yourself in OS3, please contact us using the contact email address provided in Section 14. Before we can respond to a request to exercise one or more of the rights listed above, you may be required to verify your identity. We will respond to your request within one month of receipt. Where necessary, taking into account the complexity and number of requests, we may extend this period by up to two further months; if we do, we will inform you of the extension and the reasons for it within one month of receipt.

We encourage you to contact us if you are not satisfied with how we have responded to any of your rights requests. We would, however, appreciate the chance to deal with your concerns before you approach the supervisory authority. You also have the right to lodge a complaint with your local data protection supervisory authority at any time.

9. California Privacy Rights

This Section applies to residents of California and supplements the information in the rest of this Policy. It describes your rights under the CCPA/CPRA and how to exercise them.

9.1 Notice of Collection

Although the information we collect is described in detail in Section 3 above, the categories of personal information that we have collected in the past 12 months are as follows:

• Account and registration information: email address, account identifier, password (managed by identity provider), display name, reported city/region, time zone, device display names;

• Node pairing and device environment information: Node identifier and public key, connection status, operating system and distribution, architecture, shell, hostname, OS username, rabbit agent version;

• Conversation, input and task data: chat input and instructions, uploaded files and attachments, task records, prompts and context, tool-call results, Output;

• Device-control and desktop operation data (DLAM): tool arguments and results, command output, screenshots and on-screen content (transient), input simulation records;

• File and content data accessed for tasks: content of files accessed during a task (processed transiently; original file remains on your device, though excerpts or results may appear in conversation history, Memory, or operational logs per Section 3.5);

• Memory and inferred profile data: Memory entries, embedding vectors, source conversation references;

• BYOK Model credentials and Channel pairing information: provider name, endpoint, API key (encrypted), Telegram user/bot identifiers and bot token, phone number and/or Apple ID email for Messages, and the OpenAI API key and project ID for phone calls (encrypted);

• Phone call information: recipient phone numbers you specify for outbound calls, call metadata (duration, status, timestamps), and call transcripts generated by OpenAI (see Section 3.10 for third-party sharing details);

• Log, usage and diagnostic data: IP address, browser type and version, operating system, device identifiers, access times, usage counters, operational and diagnostic logs, Node lifecycle audit logs;

• rabbit hardware (including r1) data: device identifiers, content uploaded from the device, device usage and diagnostic data;

• Information from third parties: authentication information from identity provider, messages/files from Telegram and Messages, call metadata from the telephony providers, order and shipping information;

• Personal information categories listed in the California Customer Records statute (Cal. Civ. Code § 1798.80(e)), including email address and account identifiers;

• Internet activity information: browsing history, search history, and information regarding your interaction with the Service;

• Sensitive personal information: where you include it in Input or it is accessed during a task, including financial account information, precise location, biometric data, or government identifiers;

• SOUL/persona document: profile and personality text provided by you and derived by AI, used to personalize replies;

• r1 data: when you use “export data” in OS3 settings and select “only r1 data”, your r1 data (Journal entries, Intern/Hoover sessions, photos, audio, images, documents, and artifacts, excluding conversations; earlier rabbithole data and newer r1 data are included in the same download), exported as a ZIP;

• Optional workspace scan and coding-agent import data: where you opt in, workspace scan results (which are processed like other tasks and may become part of your Memory), and selected coding-agent instructions;

• HTTP tunnel and local-proxy data: proxied HTTP/WebSocket traffic to your Node and sealed short-lived proxy tokens, streamed transparently and not persistently stored;

• Abuse-prevention and onboarding analytics: normalized email-derived bucket hash for rate limiting, and onboarding step, attempt counts and error codes.

For more detail on the information we collect, including the sources we receive information from, please review Section 3 above. In the past 12 months, we have collected and used information from all the above categories for the business and commercial purposes described in Section 4.

When we disclose personal information for a business purpose, we enter into a contract that describes the purpose and requires the recipient to both keep that personal information confidential and not use it for any purpose except performing the contract. We will not collect additional categories of personal information or use the personal information we collected for materially different, unrelated, or incompatible purposes without providing you with notice.

9.2 Right to Know and Delete

California residents have the right to delete the personal information we have collected from you and the right to know certain information about our data practices. You have the right to request the following:

• The categories of personal information we have collected about you;

• The categories of sources from which personal information was collected;

• The categories of personal information about you we disclosed for a business purpose or sold;

• The categories of third parties to whom the personal information was disclosed for a business purpose or sold;

• The business or commercial purpose for collecting or selling personal information; and

• The specific pieces of personal information we have collected about you.

In the Account section of OS3 settings, you may choose “export data” to download all your data or only your r1 data, “delete all your data” to erase all data stored under your account while keeping your account, or “delete account” to close your account, including your sign-up email (see Section 12).

To exercise any of these rights, please submit a request to our email address provided in Section 14. In the request, please specify which right you are seeking to exercise and the scope of the request. We may require further information from you to help us verify your identity and process your request. If we are unable to verify your identity or your authority to make the request, we may deny your requests to know or delete.

There are certain exceptions to the right to delete. For example, we may deny a deletion request where we need to retain the information to:

• Complete the transaction for which we collected the personal information, provide a good or service that you requested, take actions reasonably anticipated within the context of our ongoing business relationship with you, or otherwise perform our contract with you;

• Detect security incidents, protect against malicious, deceptive, fraudulent, or illegal activity, or prosecute those responsible for such activities;

• Debug products to identify and repair errors that impair existing intended functionality;

• Exercise free speech, ensure the right of another consumer to exercise their free-speech rights, or exercise another right provided for by law;

• Comply with the California Electronic Communications Privacy Act (Cal. Penal Code § 1546 et seq.);

• Engage in public or peer-reviewed scientific, historical, or statistical research in the public interest that adheres to all other applicable ethics and privacy laws, when the information’s deletion may likely render impossible or seriously impair the achievement of such research, provided you have provided informed consent;

• Enable solely internal uses that are reasonably aligned with consumer expectations based on your relationship with us; or

• Make other internal and lawful uses of that information that are compatible with the context in which you provided it.

9.3 Right to Limit Use of Sensitive Personal Information

California residents have the right to limit the use of sensitive personal information to that which is necessary to perform the service or provide the goods reasonably expected by an average consumer who requests such goods or services. Because we process sensitive personal information only when you include it in a task or it is necessary to complete a task you request, and we do not use it for other purposes, you may limit such processing by not including sensitive information in your Input or by contacting us.

9.4 Right to Correct

California residents have the right to correct inaccurate personal information that we maintain about you, considering the nature of the personal information and the purposes of the processing. You can ask OS3 in a conversation to correct information it has remembered about you (see Section 6.3). If you cannot correct inaccurate information yourself in OS3, please contact us using the details in Section 14.

9.5 Response Time and Format

Under this Section 9, we endeavor to respond to a verifiable consumer request within forty-five (45) days after we can reasonably verify your identity and authority to make the request. If we require more time, we may extend the response period by up to an additional 45 days, and we will inform you of the extension and the reason for the delay. The response we provide will also explain the reasons we cannot comply with a request, if applicable. For any request to know, we will select a format to provide your personal information that is readily usable and should allow you to transmit the information from us to another party without hindrance. We do not charge a fee to process or respond to your request unless it is excessive, repetitive, or manifestly unfounded.

9.6 Authorized Agent

You can designate an authorized agent to submit requests on your behalf. However, we will require written proof of the agent’s permission to act on your behalf, and we may still require you to verify your identity directly with us.

9.7 Right to Non-Discrimination

You have the right not to receive discriminatory treatment by us for the exercise of any of your rights under the CCPA. We will not deny you goods or services, charge you different prices or rates, provide you a different level or quality of goods or services, or suggest that you will receive a different price or rate for goods or services or a different level or quality of goods or services, because you exercised your rights under the CCPA.

9.8 Do Not Sell or Share My Personal Information

We do not sell your personal information. Where you opt in to advertising cookies, we share limited identifiers with advertising partners for cross-context behavioral advertising; you may opt out of such sharing at any time by enabling Global Privacy Control (GPC) in your browser or by emailing privacy@rabbit.tech. You can also refuse non-necessary cookies when the consent banner appears, or manage cookies through your browser settings. We honor GPC signals as an opt-out request.

9.9 Shine the Light

California’s “Shine the Light” law permits users of our Service that are California residents to request certain information regarding our disclosure of personal information to third parties for their direct marketing purposes. We do not share personal information with third parties for their direct marketing purposes. If this law applies to you and you wish to obtain further information about our sharing, please send an email to us via the contact address provided in Section 14.

10. Data Security and Retention

10.1 Data Security

We have implemented appropriate technical, administrative, and physical security measures designed to secure your personal information from accidental loss and from unauthorized access, use, alteration, and disclosure. These measures include:

• Transmission encryption: All data transmitted between the rabbit agent, Channels, and our cloud servers is encrypted using TLS;

• Storage encryption: Hosted database records are encrypted at rest (for example, AES-256);

• Envelope encryption of BYOK API keys and phone call OpenAI API keys using AWS Key Management Service, with masked display;

• Sealed, HttpOnly, Secure/SameSite first-party session cookies and short-lived, single-use pairing and proxy tokens;

• Local generation and retention of Node private keys on your device, access controls, least-privilege permissions, and audit logs of Node lifecycle actions;

• Network and application security, rate limiting, abuse monitoring, and restricted internal access on a need-to-know basis.

We regularly review our security measures to consider available new technology and methods. The safety and security of your information also depend on you. Where we have given you (or where you have chosen) a password for access to certain parts of the Service, you are responsible for keeping this password confidential and for safeguarding your API keys, paired devices, and Channels.

Unfortunately, the transmission of information via the internet is not completely secure. Although we do our best to protect your personal information, we cannot guarantee the security of your personal information transmitted to our Service. If a personal data breach occurs that is likely to result in a risk to your rights and freedoms, we will notify the competent supervisory authority without undue delay and, where required by law, notify you without undue delay.

10.2 Data Retention

We retain information for as long as necessary to provide the Service and for the other purposes set out in this Policy. The retention periods differ depending on the type of information and the purposes for which it is used. The principal periods are:

Call transcripts are retained on our servers for 30 days and then deleted, whether or not you delete your account or your data. Call audio is not stored by Rabbit; audio processed by Twilio or Photon is retained in accordance with those providers’ retention policies, and call audio or transcripts processed by OpenAI under your own OpenAI account are retained in accordance with OpenAI’s own retention policies.

Where data no longer needs to be retained, we delete, anonymize, or aggregate it. Deletion of your data or your account is subject to identity verification and to the limited exceptions described above; residual copies may persist in backups for up to 12 months until they are overwritten or deleted.

11. International Data Transfers

The Service is hosted and operated primarily in the United States, and our service providers, including third-party AI model providers selected by Rabbit, may process personal data in the United States and other jurisdictions, including countries that are not subject to an adequacy decision. If you are located in the EEA, the UK, or Switzerland, your personal data may be transferred to, stored in, or processed in countries outside your region, including the United States, where data protection laws may differ from those in your region.

When we transfer your personal data outside the EEA, the UK, or Switzerland, we ensure that appropriate safeguards are in place and that the transfer complies with applicable data protection laws, including GDPR Article 46, as follows:

• Adequacy decisions: where we transfer your personal data to a recipient in a country or territory that benefits from an adequacy decision by the European Commission (or the UK Secretary of State), we rely on that adequacy decision. This includes the EU-U.S. Data Privacy Framework and its UK Extension (and, where applicable, the Swiss-U.S. Data Privacy Framework), in each case only to the extent that Rabbit or the relevant recipient is certified thereunder and while the relevant adequacy decision remains in force;

• Standard Contractual Clauses: where no adequacy decision applies to the specific transfer, or where Rabbit or the relevant recipient is not certified under an adequacy framework, we put in place the European Commission Standard Contractual Clauses (“SCCs”) and, where applicable, the UK International Data Transfer Addendum, together with any supplementary measures (such as in-transit encryption and access minimization) required to ensure an adequate level of protection;

• Transfer impact assessments: before relying on SCCs or other safeguards for a transfer to a non-adequate country, we assess the laws and practices of the destination country to determine whether supplementary measures are needed, and we document the assessment;

• Obtaining a copy: a copy of the relevant safeguards can be requested through the contacts in Section 14.

We use Twilio and Photon to provide telephony service for OS3 phone calls. Both providers are based in the United States and may process your personal data (including called-party phone numbers, call metadata, and call audio for transport) in the United States. Your personal data processed through these providers may be stored and processed in the United States.

We use Photon to deliver Messages. When you use Messages as a Channel, the identifiers used to link the Channel (phone number and/or Apple ID email), the messages and attachments you send and receive through Messages, and delivery status are relayed between you and OS3 by this provider. Photon is a company based in the United States. Your personal data processed through this provider may be stored and processed in the United States.

For BYOK Models, Skills, and user-requested destinations, you may direct transfers to providers or locations of your choosing; you are responsible for assessing those providers, and the transfer is governed by your agreement with that provider.

12. Account Deletion and Data After Termination

In the Account section of OS3 settings, you can delete all your data, or delete your account. Before deleting, you may wish to export a copy. Use “export data” and select “all data” to download everything in OS3 and on r1 in one download, or select “only r1 data” to download your r1 data (including intern sessions, journal, photos, audio, and artifacts, but not conversations; earlier rabbithole data and newer r1 data are included in the same download). You can also ask OS3 in a conversation to export your conversation history and Memory (see Section 8.4).

Choose “delete all your data” to permanently delete your conversations, Memory and its embeddings, personality, name, skills, schedules, connected computers, saved keys, and r1 data, and to unpair your r1. Your account stays, and you will start setup again afterwards. This cannot be undone.

Choose “delete account” to permanently delete your account and all your data across OS3 and r1, including your sign-up email. You will be logged out and will not be able to sign in to this account again. This cannot be undone.

For either deletion: settings and files stored on your own computers (for example, MCP servers and their tokens, and installed skills) are not affected; the rabbit agent on each connected computer will uninstall itself shortly; and to erase data on the r1 device itself, factory reset it before deleting.

On either deletion, we delete your personal information from our active systems — including conversation history, Memory (with the embeddings used to index it), and your SOUL document — within 30 days, subject to the retention periods and exceptions in Section 10.2. On “delete account”, we also stop providing the Service to that account.

Please note that deleting your data or your account removes Rabbit’s copy of your API keys but does not revoke or disable those keys on the model provider’s side; you must revoke them directly with the relevant provider.

Upon termination or expiration of your access to the Service, your license ends and you must cease use and uninstall the rabbit agent. Uninstalling the rabbit agent removes the local application and its local configuration directory but does not by itself delete your cloud account or conversation history; you must delete your data or your account separately. After termination, sections of this Policy that by their nature survive will survive, including our rights to retain information as required by law and to enforce our Terms.

13. Changes to This Privacy Policy

We may update this Privacy Policy as our Service and applicable law evolve by posting a revised version with a new “Last Updated” date. For material changes, we will provide reasonable notice, which may include posting the revised Privacy Policy on our website, updating the “Last Updated” date, in-Service notification, or email to your account email. For consumers in the EU/EEA/UK, material changes will not be imposed retroactively; we will notify you through in-Service notification or other means that directly reach you, and you will be given a reasonable opportunity to cancel your account or withdraw consent before they take effect; continued use after the effective date constitutes acceptance where permitted by mandatory law. If you do not agree with a change, you must stop using the Service.

14. Contact Information, Data Protection Officer and EU/UK Representative

Rabbit Inc. is the controller and responsible for your personal data.

Full name of legal entity: Rabbit Inc.

Postal address: 1626 Montana Avenue, Suite 162, Santa Monica, California 90403, USA

For privacy questions, requests, complaints, or general support, to reach our Data Protection Officer: privacy@rabbit.tech

EU / UK Representative. Pursuant to Article 27 GDPR (and the UK GDPR), our representative in the European Union/United Kingdom is:

Dr. Jörg Brettschneider, Brettschneider Rechtsanwaltsgesellschaft mbH,
Alter Wall 32, 20457 Hamburg, Germany.
Phone: +49 40 3346 64190

Fax: +49 40 3567 4881
Email: ra-gdpr@brettschneider.law

You have the right to make a complaint at any time to the supervisory authority for data protection issues in your jurisdiction. We would, however, appreciate the chance to deal with your concerns before you approach the supervisory authority.