← AI Awareness Fundamentals
Module 1: Understand the Tool You Are Talking To
Why Fluent Is Not the Same as True Capability Has Boundaries The Safe Input Decision
Module 2: Know the Service and Its Reach
Who Is Behind This AI Service? Where Prompts, Files, and Outputs Can Travel AI Agents, Connectors, and Hidden Instructions
Module 3: Verify, Decide, and Recover
Verify Claims, Citations, Code, and Summaries Deepfakes, Synthetic Media, and Persuasive Output A Responsible Personal AI Practice
Module 3: Verify, Decide, and Recover

A Responsible Personal AI Practice

Combine safe inputs, provider evidence, verification, human judgment, disclosure, and recovery into one repeatable practice.

In this lesson, you will learn to:

  • Run a complete pre-use and pre-action check.
  • Respond to sensitive disclosure, harmful output, or unintended agent action.

A Responsible Personal AI Practice

A capstone lesson for selecting use cases, retaining accountability, recognizing incidents, and improving how you use AI over time.

You remain the decision owner

Before using AI, define the task, affected people, information boundary, consequence of error, and evidence needed. Choose the exact service only after checking its provider and data behavior. Transform or minimize input, ask for a draft, and keep the original source available. Before using the result, verify the claims and consider whether readers need disclosure.

Human review is meaningful only when the reviewer can understand and reject the proposed action. A person clicking approve without seeing recipients, data, evidence, and consequences is ceremonial. Slow down for decisions affecting health, livelihood, rights, money, access, public reputation, or safety. Sometimes the responsible choice is not to use AI because the task, service, or evidence is a poor fit.

The human stays in the decision cockpitA cockpit places purpose, input, provider, evidence, and consequence around the final human decision.LEARN IT VISUALLYThe human stays in the decision cockpitHuman decisionPurposeSafe inputProviderEvidence
The human stays in the decision cockpit — A cockpit places purpose, input, provider, evidence, and consequence around the final human decision.

Treat AI mistakes and disclosures as incidents

If sensitive data enters an unsuitable service, stop adding information and record the service, account, time, feature, data categories, links or connectors, and deletion actions. Do not paste the same data into another chatbot to ask whether it is sensitive. At work or school, contact the relevant security, privacy, or support route. Personally, use provider support and protect any exposed credentials or identities.

If an agent takes an unintended action, disable its connectors or account access, preserve the activity log, reverse the action where safe, and inspect downstream effects. If generated misinformation was published, correct it visibly and notify affected people. If a deepfake impersonation caused payment or account disclosure, contact the service or financial provider through a known route and preserve evidence.

An AI incident should improve the next cycleA spiral moves from detect to contain, assess, correct, and improve without hiding the mistake.LEARN IT VISUALLYAn AI incident should improve the next cycleDetectContainAssessCorrectImprove
An AI incident should improve the next cycle — A spiral moves from detect to contain, assess, correct, and improve without hiding the mistake.

Keep learning because the service changes

Models, defaults, integrations, laws, and threat techniques change quickly. Revisit your provider evidence and connected apps. Test whether saved memory, chat history, and training controls still behave as expected. Review important uses when a new model or feature appears rather than assuming the earlier decision transfers automatically.

Current European Commission guidance explains that AI literacy measures should reflect the person’s knowledge, experience, context, and the people affected; it does not prescribe one universal individual score. That principle works beyond legal compliance: learn enough for the actual tools and consequences you face. A good personal practice is observable and repeatable—know the service, protect the input, question the output, control the action, and learn from the result.

Resources

← Previous: Deepfakes, Synthetic Media, and Persuasive Output