Conversational Privacy

Audience

Individuals using consumer chatbots, families using shared subscriptions or devices, employees and students using approved AI tools, and small organizations without dedicated privacy or security teams. No technical, legal, security, or privacy background is required. Learners may use printed scenarios or observe a demonstration; no chatbot account is required.

Purpose

Help learners transfer familiar privacy instincts into chatbot conversations so they can choose the setting, information, and level of detail with care.

What Do You Need

• Format: individual, pair, small group, observation, or facilitated demonstration.
• Tool: a chatbot is optional. Printed cards and fictional scenarios can support the full lesson.
• Materials: one supplied privacy scenario and paper for optional notes.
• Recommended preparation: 🧩 Module · AI Without the Hype and 🧩 Module · What AI Is and Is Not, the module's stated dependencies, introduce basic AI expectations and human responsibility. Do not bring private records or reveal real account details.

Learning Outcomes

By the end of this module, learners should be able to:

• Explain why a personal or intimate conversation may still not be private.
• Notice how conversational interfaces can reduce the friction to sharing.
• Consider information risk by asking what could be done with the information.
• Distinguish useful context from unnecessary identifying precision.
• Recognize how several broad details can identify someone when combined.
• Protect information that belongs partly or entirely to another person.
• Pause before sharing while continuing to use approved tools productively.

Intimate Does Not Mean Private

A chatbot can respond in a warm, patient, and personal way. That feeling does not establish privacy. Treat the conversation according to where the information can travel, who or what may gain access under the product's current conditions, and what the information could enable—not according to how private the exchange feels.

This does not mean every conversation is being watched by a person. It means the interface's conversational feeling is not enough evidence for a disclosure decision.

Conversation Lowers Sharing Friction

Disclosure often grows one ordinary turn at a time. A first request may be harmless. A follow-up adds a name, location, date, workplace, diagnosis, account detail, or story about another person. None of the turns feels like a dramatic decision, but together they can create a revealing record.

Notice the accumulation. A few blurry details can become a sharp picture when combined.

What Could Be Done With It?

Instead of trying to memorize one universal list of sensitive information, ask what the information could enable if it traveled farther than expected:

• Access to an account, device, place, or service.
• Theft, impersonation, targeting, pressure, embarrassment, or discrimination.
• Harm to a student, family member, client, employee, patient, or organization.
• A connection between details that identifies a person or situation.

The possible harm, setting, audience, and purpose all matter. Privacy decisions are contextual, and privacy is a shared responsibility among people, organizations, and platforms—not a burden learners carry alone.

Use the Least Precise Useful Information

Give enough information to support the task, but not enough to pinpoint the person or expose an unnecessary record. You might replace a full name with a role, an exact date with a range, an address with a region, or a real case with a fictional equivalent.

Precision is not always wrong. Exact detail may be necessary for an approved task, emergency, professional process, or authoritative record. When it is necessary, use the right system and permission rather than assuming a general chatbot is the right setting.

Protect Other People

Being allowed to tell a story is not the same as having permission to place every detail into a tool. Before sharing information about another person, ask:

• Does the task require their information?
• Could I remove or generalize identifying details?
• Do I have permission and an approved setting for this use?
• Would the person reasonably understand how their information is being used?

Give extra care to information about children, students, patients, clients, employees, family members, and people who cannot meaningfully consent.

Accounts, Devices, Histories, and Deletion

Shared accounts and devices can blur who can see a conversation and whose identity shapes it. Saved histories, memory features, uploaded files, connected services, exports, notifications, and workplace or school administration can change the practical privacy boundary. Current behavior varies by product, plan, setting, and time.

Deleting a visible conversation may reduce future exposure, but it does not reliably undo disclosure, copies, exports, screenshots, prior access, or downstream use. Check current product and organizational rules for the setting you are using.

The Privacy Pause

Before sharing, ask:

• What am I sharing?
• What could be done with it?
• Why is it needed?
• Can I reduce the precision?
• Where is the conversation happening?
• Who else could gain access?
• Does the information belong only to me?
• Would I be comfortable if the conversation traveled farther than expected?

The pause is a decision aid, not a guarantee. If the stakes are high or the rules are unclear, stop and use an approved person, policy, or system.

Practice

Use What Does the Tool Need? (legacy Learning Lab), Beta, to sort a fictional packet and write a minimum-necessary brief. Its Lab Notebook (legacy Learning Lab) carries the activity and an optional place to record your decisions.

Start with The Restaurant Test (legacy Learning Lab) to compare how the same disclosure feels in different settings. Then use Reduce the Precision (legacy Learning Lab) to rewrite fictional requests while preserving their useful goal. The Disclosure Ladder (legacy Learning Lab) makes gradual accumulation visible.

Reflection

• Which detail changed the risk of the scenario?
• What information supported the task, and what merely identified someone?
• Which information belonged to another person?
• What safer wording, setting, or route preserved the useful goal?
• What would make you stop and ask for help?

Where to Go Next

🧩 Module · Relationships with Conversational AI examines why a system can feel trustworthy, caring, dependable, or relationship-like. Conversational Privacy stays focused on what belongs in a conversation under its actual conditions.

Tool Boundary

No chatbot account or paid feature is required. Product settings, memory, retention, deletion, administration, and connected-service behavior change. Use current product documentation and organizational policy for real decisions. This module is practical education, not legal, privacy, security, medical, workplace, or school-policy advice.

Privacy Boundary

Use fictional, synthetic, public, or heavily sanitized examples. Do not enter passwords, account numbers, identity documents, medical details, school records, client or employee files, intimate details, private family conflicts, confidential business information, or another person's information without authorization. When real harm is possible, use the approved channel and a qualified person.