Frontier — Privacy Policy
1. Introduction
This Privacy Policy explains how Parallel Frontier Labs Inc. (“Parallel Frontier,” “we,” “us,” or “our”) collects, uses, discloses, and protects information in connection with the Frontier desktop application, the inference and control-plane services that support it, and the parallelfrontier.io website (together, the “Services”).
Frontier is a business product. Most people who use Frontier do so through an organization that has an account with us. When we process data on behalf of a business customer as part of providing the Services, we act as a processor (or “service provider”), and that processing is governed by our agreement with that customer and, when in place, our Data Processing Addendum (“DPA”), rather than by this Policy. In case of conflict, the customer agreement and any applicable DPA will control.
This Policy describes our practices where we act as a controller — account administration, billing, website visitors, and direct communications with us.
If you use Frontier through your organization, please also review your organization’s own privacy notices. Your organization controls its account and the data it processes through the Services.
2. A note for financial-services and other regulated customers
Financial advisory firms and other regulated customers may submit NPI or other regulated personal data to Frontier. When they do, we handle that data as a service provider under the customer’s instructions, subject to our customer agreement and, when in place, a DPA. Customers remain responsible for deciding whether and how to use Frontier with regulated data and for their own SEC, Regulation S-P, books-and-records, privacy, consent, retention, supervisory, and vendor-management obligations unless Parallel Frontier separately agrees otherwise. We provide reasonable information about our handling of customer data to support those obligations, but this Policy is not a certification that Parallel Frontier is itself regulated as a financial institution.
Three answers a compliance officer usually needs first:
| Question | Answer |
|---|---|
| Where does our data rest? | Depending on your deployment model, task transcripts, execution traces, and saved files are stored in your Customer Cloud/BYOC environment (where offered) or in Hosted PFL Cloud storage scoped to your account. See §3. |
| Who at Parallel Frontier can see it? | Access is least-privilege and logged. Support personnel access customer content only with your authorization, for a defined session, as described in the customer agreement and, when in place, a DPA. |
| When do you tell us if something goes wrong? | We notify affected customers without undue delay after becoming aware of a security incident involving their data and, where required by the applicable customer agreement or DPA, no later than forty-eight (48) hours after awareness. |
We do not use customer content to train models of our own, and §6 explains the limits of provider-specific retention and training commitments.
3. What we collect, and exactly where it goes
Because Frontier is an autonomous agent that reads and writes files in the environment you authorize, the most useful thing we can tell you is a practical map of what may leave that environment, what we store, and what generally stays under your control.
3.1 What generally stays in your authorized environment
Your local task database and Agent working state.
Your authentication tokens and optional configuration files.
The Agent’s persistent browser profile, including any sessions you sign into through it.
Credentials for any third-party AI subscription you connect in local mode — these are read and used on your machine. We store only a non-secret indicator that a subscription is available.
3.2 What passes through us but is not retained by us
To perform a task, the Agent transmits task inputs — including prompt contents, files it reads from folders you attach, target URLs and search queries, and, if you use dictation, microphone audio — through our inference gateway to the AI, web-data, and transcription providers listed on the Subprocessors page.
At the gateway we do not persist these request payloads beyond the transient processing needed to route and deliver each request. We retain only operational metadata: token counts, the provider used, timestamps, status, and computed cost.
Retention at the provider is a separate question, answered in §6.
3.3 What we store
| What | Where | How long |
|---|---|---|
| Task transcripts, execution traces, files the Agent produces or that you choose to save | Customer Cloud/BYOC environment (where offered) or Hosted PFL Cloud storage scoped to your account, accessed with short-lived credentials | As provided in the customer agreement, service configuration, and deletion or export instructions, subject to applicable law and regulated-recordkeeping obligations |
| Task metadata — title, status, model used, timestamps, token counts, cost | Application database, United States | Up to 1 year, unless deleted earlier on request or a different period is required by the customer agreement, applicable law, or regulated-recordkeeping obligations |
| Usage and billing ledger | Application database | Up to 1 year, unless deleted earlier on request or a different period is required by the customer agreement, applicable law, or regulated-recordkeeping obligations |
| Account and identity — name, email, authentication identifier, invited addresses and roles | Application database + Firebase Authentication | Term of account, then up to 1 year unless deleted earlier on request or a different period is required by the customer agreement, applicable law, or regulated-recordkeeping obligations |
| Organization and workspace — names, settings, domain allowlist, memberships, entitlements | Application database | Term of account, then up to 1 year unless deleted earlier on request or a different period is required by the customer agreement, applicable law, or regulated-recordkeeping obligations |
| Saved automations — prompt text, working directories, attached-file references, settings | Application database | Until you delete them or as provided in the customer agreement, service configuration, and deletion or export instructions, subject to applicable law and regulated-recordkeeping obligations |
| Server and usage logs | Hosting infrastructure | Up to 1 year |
| Communications with us | Email and support tooling | 1 year, unless earlier deleted on request or as required by law |
You can delete individual tasks and automations, and you may request deletion of records associated with your account. Usage and billing ledger entries — which may include a snapshot of a task’s title — are retained for the billing-records period even if the underlying task is deleted. When an organization’s account is deleted, database records and stored task content are deleted or handled under the customer agreement and applicable legal requirements, subject to security, fraud and abuse prevention, backups, billing and tax records, disputes, legal holds, and any regulated-recordkeeping obligations that apply to the customer.
3.4 What we do not currently collect
We do not currently embed third-party advertising, analytics, or behavioral-tracking software development kits in the Frontier desktop app or backend. We do not knowingly collect special categories of data for our own purposes, and we do not sell personal information.
(If we add error-reporting, analytics, or similar tooling, we will update this Policy and the Subprocessors page.)
4. Website and cookies
Our current marketing website at parallelfrontier.io does not set its own cookies and does not run analytics or behavioral-tracking scripts. It may load third-party assets — such as web fonts, images, video, or other content-delivery services — which, like any web request, can transmit your IP address and browser user-agent to those providers so they can deliver the asset.
If we add analytics or other cookies in the future, we will update this Policy and provide any legally required notice or consent mechanism. Our website does not respond to “Do Not Track” browser signals. Because we do not sell or share personal information or engage in cross-context behavioral advertising, opt-out preference signals such as the Global Privacy Control do not change how we process personal information.
5. How we use information
We use information to: provide, operate, maintain, and secure the Services; authenticate users and manage accounts, workspaces, memberships, and invitations; route your requests to the providers needed to complete your tasks; measure usage and calculate and collect fees; provide support and respond to your communications; monitor for, investigate, and prevent fraud, abuse, security incidents, and violations of our Terms; maintain business records; and comply with legal obligations and enforce our agreements. Where the EU or UK GDPR or similar laws apply to processing we control, our legal bases are performance of a contract, our legitimate interests (such as securing, operating, and improving the Services), compliance with legal obligations, and, where applicable, consent.
Where we act as a service provider for a business customer, we use that customer’s content and personal data only to provide the Services and on the customer’s documented instructions.
6. Training and retention at AI providers — read this carefully
Content can reach an AI provider in different ways, and the applicable retention and training protections depend on the route.
Route 1 — Providers we contract with, open-weights models
Together AI and Fireworks AI currently serve open-weights models to us on their own infrastructure; the Subprocessors page is the authoritative, current list. Both providers operate with zero data retention for our inference traffic: prompts and outputs are processed in memory to serve each request and are not stored by the provider, and neither provider uses our customers’ content to train models. Prompt storage, data-sharing, and model-training options are disabled at the organization level on our accounts, and we have verified those settings. These protections reflect each provider’s published terms and our account configuration; our Terms state the no-training and retention commitments that apply to providers we contract with, and we will reflect any changes on the Subprocessors page.
Route 2 — Providers we contract with, frontier models for specific features
Certain features route to frontier providers — currently OpenAI, Anthropic, Google, and xAI — under our accounts and their business API terms. Under those terms, these providers do not use API inputs or outputs to train their models, and retain them only for a limited period (typically up to 30 days) for abuse monitoring and legal compliance. We have not established zero-data-retention arrangements with these providers. Customers intending to submit NPI or other regulated personal data to features that route to these providers should first confirm these protections satisfy their obligations, or contact us about stricter arrangements.
Connecting your own AI subscription does not move all processing to that provider relationship. The Subprocessors page identifies the providers we engage directly, including hosted sandbox providers where applicable.
Request classification. TypeSafe AI classifies requests so they are routed to the right agents, and helps choose which attached files a task needs. It receives the text of the request and basic information about the task and its attached files, such as file names and sizes. Never file contents. TypeSafe AI is listed on the Subprocessors page. We have not established a zero-data-retention arrangement with TypeSafe AI.
Route 3 — Your own AI subscription, connected by you
If you connect your firm’s own Anthropic, OpenAI, Google, or xAI subscription, that work — other than the Route 2 features above — runs under your provider relationship.
Retention and training behavior is governed by your agreement with that provider — not by ours. We are not a party to it and cannot warrant it.
Consumer and free tiers may not provide zero-retention or no-training protection and may permit human review. Enterprise agreements may provide stronger protections.
We do not verify which tier you connect. If you intend to submit NPI or other regulated data through a connected subscription, confirm your agreement with that provider supports it before connecting.
In all cases
We will not use Customer Content or Output (each as defined in our Terms & Conditions) to train, fine-tune, develop, or improve any AI model of ours. We may use aggregated or de-identified operational data — token counts, error rates — that does not identify you or your organization, to monitor, secure, and improve the reliability of the Services.
7. How the Agent accesses and processes your data
Frontier is an autonomous coding agent, so how it handles data matters.
You control which folders and files the Agent can access. It is designed to read only the files needed for a task, and relevant contents may be transmitted to AI providers as described in §3 and §6. The Agent can write files back to your device when you direct it to, and depending on your deployment model, runs code in an isolated sandbox on your device, in your cloud/BYOC environment (where offered), or in Parallel Frontier’s hosted cloud environment.
The Agent’s built-in browser. If you enable it, the Agent operates a browser with a persistent profile that can retain sessions you sign into. That profile can function like a credential store in the environment where it runs. You are responsible for securing it — including clearing sessions when appropriate and controlling who can access the environment.
You remain responsible for the data you direct the Agent to access and the actions you authorize. See the Terms & Conditions and the Acceptable Use Policy.
8. How we share information
We do not sell your personal information. We share only as follows:
Service providers and subprocessors. Vendors that support delivery of the Services — cloud hosting and infrastructure, sandbox execution, database and storage, authentication, AI model inference, request classification, web-search and web-data, speech-to-text, email delivery, security and abuse monitoring, customer support, and payment and billing. We maintain a Subprocessors page that identifies key vendors and their functions. Our customer terms describe applicable subprocessor commitments, including notice of changes.
Business account administrators. If you use Frontier through an organization, your workspace administrators can access and manage your account and its data.
Legal, safety, and protection of rights. Where we believe it necessary to comply with law or legal process, enforce our agreements, or protect the rights, safety, security, and property of you, us, or others.
Corporate transactions. In a merger, acquisition, financing, reorganization, or sale of assets, subject to the protections in this Policy.
Professional advisors and affiliates, under obligations of confidentiality.
Where required, we seek to impose contractual data-protection obligations on parties with whom we share personal data, and we are continuing to roll out updated enterprise data-protection terms across our vendor and subprocessor base.
9. International data transfers
For Hosted PFL Cloud, customer content is generally processed and stored in the United States unless the applicable customer terms or product documentation say otherwise. For Customer Cloud/BYOC deployments (where offered), data location depends on the customer’s selected environment and configuration.
Where we transfer personal data from the European Economic Area, the United Kingdom, or Switzerland, we rely, where required, on appropriate safeguards, which may include the European Commission’s Standard Contractual Clauses and, where applicable, the UK Addendum. Because our subprocessors are U.S.-controlled entities, data may be subject to U.S. law regardless of where it is stored.
10. Data retention
Where we act as a service provider, we retain customer content as provided in the customer agreement, service configuration, and deletion or export instructions, subject to applicable law, regulated-recordkeeping obligations, security, fraud and abuse prevention, backups, billing and tax records, disputes, legal holds, and customer-directed retention requirements.
For data we control, retention periods are stated in the table at §3.3.
11. Security
We maintain administrative, technical, and physical safeguards designed to protect information appropriate to its sensitivity. Safeguards vary by deployment model and are designed to include measures such as encryption of data in transit and at rest, tenant isolation, least-privilege access controls, scoped credentials, and controlled access provisioning.
Additional security information may be made available to business customers and prospects under NDA. We are pursuing SOC 2 readiness but do not currently hold SOC 2, ISO 27001, PCI, HIPAA, FedRAMP, or equivalent certifications or attestations.
No method of transmission or storage is completely secure, and we cannot guarantee absolute security.
12. Your rights and choices
Depending on where you are and your relationship with us, you may have rights to access, correct, delete, or port your personal information, to object to or restrict certain processing, and to withdraw consent.
Where we process personal data on behalf of a business customer, we refer your request to that customer and support them in responding. Direct such requests to your organization’s administrator.
For data we control, contact [email protected].
How we handle requests. We may ask for information to verify your identity and will respond within the time required by applicable law. If we deny a request in whole or part, we will explain why and tell you about any available appeal rights. You may also lodge a complaint with a supervisory authority.
We do not sell or “share” personal information for cross-context behavioral advertising, we do not use it for targeted advertising, and we do not use or disclose sensitive personal information beyond the purposes permitted without offering a limitation right. We will not discriminate against you for exercising your rights.
U.S. state privacy laws. Residents of states with comprehensive privacy laws — including California, Virginia, Colorado, Connecticut, and others as they take effect — have the rights described above with respect to personal information we control, subject to statutory exceptions. California residents may designate an authorized agent. Certain financial data that is subject to GLBA and its implementing rules may be exempt from some state-law requirements.
Notice at Collection (California)
| Category collected | Purpose | Sold or shared? | Retention |
|---|---|---|---|
| Identifiers — name, email, account and authentication identifiers | Account administration, authentication, support, billing | No | See §3.3 |
| Commercial information — plan, entitlement, usage and billing records | Billing, dispute resolution, business records | No | See §3.3 |
| Internet or network activity — server and usage logs | Security, reliability, abuse prevention | No | 1 year |
| Audio — dictation, if you use it | Transcription to perform your task | No | Not retained by us; see §3.2 |
| Professional information — role, workspace membership | Access control | No | See §3.3 |
| Content you submit through the Agent | Performing the task you direct | No | See §3.2–3.3 |
We do not use or disclose sensitive personal information for purposes requiring a right to limit.
13. Children
The Services are business tools intended for use by organizations and are not directed to children. Our Terms require all users to be at least 18 years old. We do not knowingly collect personal information from anyone under 18. If you believe a minor has provided us personal information, contact us and we will take appropriate steps to delete it.
14. Changes to this Policy
We may update this Policy. If we make material changes, we will provide notice by updating the “Last updated” date and, where appropriate, by additional notice in-product or by email. For material changes affecting business customers, we provide the notice period in the applicable agreement. Continued use after an update takes effect constitutes acceptance.
Version history
| Version | Date | Summary of changes |
|---|---|---|
| 1.0 | August 23, 2026 | Initial publication |
| 1.1 | September 24, 2026 | Added request classification by TypeSafe AI (§6, §8) |
| 1.2 | September 24, 2026 | Simplified the TypeSafe AI description; it also helps choose which attached files a task needs. (§6) |
15. Contact us
Parallel Frontier Labs Inc. 2810 N Church St STE 89140 · Wilmington, DE 19802, United States
General: [email protected] Privacy and data protection: [email protected] Security: [email protected]
Privacy contact / responsible person: Parallel Frontier Labs Inc., Privacy Team
Related
Subprocessors · Acceptable Use Policy · Security Exhibit