How to Configure Perplexity Personal Computer Privacy and Data Controls

Perplexity Personal Computer is an always-on AI agent that runs in a secure environment connected to your local apps, files, and Perplexity's cloud servers. Unlike the standard Perplexity web app where you type a query and get a response, Personal Computer works continuously in the background. It can browse the web, read your local files, execute multi-step tasks, and operate across sessions without you needing to babysit every action.

That autonomy is the whole point, but it also raises real privacy questions. An agent with persistent local file access needs stronger guardrails than a chat window. Perplexity addresses this with three layers of control: device-level safeguards built into Personal Computer itself, account-level privacy settings that apply across all Perplexity products, and enterprise controls for organizations that need tighter governance.

The rest of this guide covers each layer and how to configure it.

Personal Computer ships with three built-in safeguards that apply specifically to the always-on agent environment. These are not optional settings you need to enable. They are active by default.

Emergency stop

Perplexity describes Personal Computer as running "in a secure environment" with "immediate control" over agent operations. You can halt all Personal Computer activity at any time. If the agent starts doing something unexpected or you want to pause operations, this gives you instant control without waiting for the current task to finish.

Approval gates for sensitive actions

Not every action runs autonomously. Personal Computer requires explicit user approval before executing sensitive operations. The specifics of what counts as "sensitive" are defined by Perplexity's system, but the pattern is clear: the agent proposes an action, you approve or reject it, and only then does it execute. This prevents the agent from making irreversible changes to your files or accounts without your knowledge.

Audit trail

Every session generates a complete audit trail. Every action the agent takes on your behalf is logged, giving you a full record of what happened, when, and in what order. If something goes wrong, you can trace back through the log to understand exactly what the agent did.

These three controls work together. The audit trail shows you what happened, approval gates prevent unwanted actions before they happen, and the emergency stop is the brake for everything else.

Beyond the Personal Computer-specific controls, your Perplexity account has several privacy settings that affect how your data is collected, stored, and used across all Perplexity products, including Personal Computer.

Disable AI Data Retention By default,

Perplexity uses your searches and responses to improve its AI models. Free, Pro, and Max plan users can opt out:

1. Open your Perplexity account settings
2. Navigate to the Preferences tab
3. Toggle off "AI data retention"

With this toggle off, Perplexity will not use your queries or the answers you receive to train or improve AI models going forward. One important caveat: opting out applies only to future data. Anything already used in training before you flipped the switch may not be retroactively removed.

If you use Perplexity without logging in, your inputs and outputs are collected for AI training by default with no way to opt out. Always log in before using Personal Computer or any other Perplexity product if you care about data retention controls.

Use Incognito Mode for Sensitive Queries

Perplexity offers an incognito mode that prevents conversations from being saved to your library. You can activate it by clicking your profile icon and selecting "Incognito," or by pressing Cmd/Ctrl + semicolon on the web.

Incognito conversations expire after approximately 24 hours and are never stored in your account history. This is useful for one-off sensitive queries where you do not want a permanent record. Note that incognito mode is a feature of the Perplexity web and app interface. Whether it applies to Personal Computer's autonomous sessions depends on how you initiate tasks.

Global Privacy Control (GPC) Signals

Perplexity recognizes GPC signals broadcast from your web browser as a valid opt-out request where required by applicable law. If your browser sends a GPC signal, Perplexity will treat it as an opt-out of targeted advertising for that browser session.

There are two limitations to know about. First, the GPC signal applies only to your current browser, not across devices or sessions. Second, Perplexity does not recognize the older "Do Not Track" signal. If there is a conflict between a GPC signal and a manual privacy choice you have made on the Perplexity website, your manual choice takes priority.

Give Your AI Agent a Workspace That Outlasts the Session

Personal Computer generates output. Fast.io gives it a permanent home with audit trails, granular permissions, and built-in AI search. 50 GB free, no credit card. Built for perplexity personal computer privacy data control settings workflows.

Sign Up Free

Perplexity applies different retention periods depending on the type of data and your account tier. Understanding these timelines helps you know what is kept and for how long.

Account data stays active as long as your account exists. If you delete your account, Perplexity removes your data within 30 days.

Search history and chats are retained until you manually delete them. They do not auto-expire for standard accounts, so if you want them gone, you need to go into your history and remove them yourself.

Uploaded files and images are auto-deleted after 30 days for regular users. Enterprise users get a shorter window of 7 days.

Incognito mode conversations expire after approximately 24 hours and are not stored in your account history at all.

Sonar API requests follow a strict zero data retention policy. Perplexity does not store any prompts or responses sent through the API, and API data is never used for model training. If you are building automations on top of Perplexity, the API path is the most privacy-conservative option.

For Personal Computer specifically, Perplexity has not published separate retention timelines for agent session data versus standard search data. The audit trail logs your agent's actions, but Perplexity's public documentation does not yet specify how long those logs are retained or whether they fall under the same 30-day file deletion policy. If this matters for your use case, contact Perplexity support for clarification before granting the agent access to sensitive files.

Organizations using Perplexity Enterprise get a stronger set of privacy defaults and additional administrative controls.

No AI training on enterprise data. Enterprise data is explicitly excluded from AI model training. There is no toggle to manage because the protection is on by default and cannot be overridden. This is the biggest difference between enterprise and individual plans.

Workspace-level access restrictions. Enterprise admins can limit data retention settings, restrict access at the workspace level, and manage sharing permissions within the organization. This gives IT teams control over who can see what and how long data persists.

enterprise security standards compliance. Perplexity holds enterprise security standards certification, which covers security, availability, processing integrity, confidentiality, and privacy. This is relevant for organizations that need to demonstrate vendor compliance to their own auditors.

SAML single sign-on and centralized audit logs. Enterprise deployments support SAML SSO for identity management and centralized audit logging for compliance reporting.

For teams deploying Personal Computer at scale, the enterprise controls layer on top of the device-level safeguards. The emergency stop and approval gates still apply per device, but the admin also gets organizational visibility into how agents are being used across the team.

Personal Computer generates files, research summaries, and task outputs that you may need to share with colleagues or hand off to other tools. Perplexity's built-in storage has its own retention limits, and Personal Computer's local file access is tied to the device it runs on. If you need persistent, shareable storage for what your agent produces, you will want an external workspace.

Fast.io works well for this because it is built for the agent-to-human handoff pattern. An agent can create workspaces, upload files, and organize outputs using the Fast.io MCP server or REST API. When the work is ready for human review, the agent transfers ownership to a team member who picks up where the agent left off.

A few features that matter for privacy-conscious agent workflows:

• Granular permissions at the org, workspace, folder, and file level, so you control exactly who sees what
• Audit trails that log every action, complementing Personal Computer's own session logs
• File versioning so you can track changes over time and roll back if needed
• Intelligence Mode that auto-indexes uploaded files for semantic search and AI-powered Q&A with citations

The free agent plan includes 50 GB of storage, 5,000 monthly credits, and 5 workspaces with no credit card required. If your Personal Computer agent produces output that needs to outlast a single session or reach people beyond your own Mac, this is a clean way to handle it.

You can also use Fast.io's URL Import to pull files from Google Drive, OneDrive, Box, or Dropbox directly into a workspace without downloading them to your local machine first. This keeps your Personal Computer's local storage clean while centralizing everything in one searchable location.