Verify Before You Act

Your objective

By the end of this lesson, you will be able to pause when a claim sounds urgent or consequential, then verify it without relying on the message, video, or AI answer that first brought it to you.

AI can help you organize questions and identify what to check. It is not proof that a current claim is true.

m7-l1-objective

See It: A convincing claim with a deadline

Jordan receives a group-chat video that appears to show a city official saying water rates will double tomorrow unless residents enroll in a relief program today. The caption includes a link that looks official at a glance.

A familiar face, clear video, and an urgent deadline can make a message feel settled before it has been checked. The video could be edited, miscaptioned, impersonated, or generated. The claim could also simply be wrong.

m7-l1-scenario

The water-bill video

What should Jordan treat as unproven, and what source could Jordan check without using the video link?

Understand It: Plausible is not verified

A hallucination is an AI response that presents incorrect or invented information as if it were reliable. It may include specific dates, names, explanations, or even source-like details. Confidence is not a quality check.

Misinformation can spread through an honest mistake or through deliberate deception. A post may quote a real source inaccurately, use an old image in a new context, or imitate a trusted person or organization.

Do not use AI as your only authority for information that could affect money, safety, health, legal rights, a deadline, or another important decision. For medical, legal, or financial matters, AI can help you understand language and prepare questions; a qualified professional or official source must guide the decision.

m7-l1-understand

What exactly is the claim, and when is it supposed to be true?

Who has direct authority to confirm it?

Can I reach that source by a web address, app, phone number, or contact I already know is legitimate?

Does a second independent, trustworthy source agree?

Try It: Ask AI to show its limits, not to replace a check

m7-l1-conversation

The first AI answer sounds useful but supplies no evidence. The refined follow-up asks for a verifiable source and gives the AI permission to say it does not know.

A safer follow-up after an unsupported answer

Learner: A video says my city's water rates will double tomorrow unless I enroll today. Is that true?

AI: Yes. The city approved a new surcharge starting tomorrow, and residents should enroll before the deadline.

Learner: Which official notice confirms that? Give the direct official city page, and clearly say if you cannot verify it. Do not guess.

AI: I cannot confirm that claim from an official current notice. My earlier answer was unsupported and should not be used. Check the city government's website by typing its known address yourself, or call the number on a prior bill.

Improve It: Separate the claim from the channel that delivered it

m7-l1-demonstration

Pause. Urgency is a cue to slow down, not a reason to act faster.

Check the source outside the message: type a known address, use a saved official app, or call a number from a trusted record.

Check the claim, date, and action requested. A genuine announcement should be findable without relying on the original post.

If the evidence remains unclear, do not act yet. Ask the appropriate organization or qualified professional.

First request: Open the video link, search the claim, and accept the first convincing answer.

Improved request: Write down the claim. Do not use the post's link or phone number. Open a known official site or app yourself, use contact information from a bill or card, and compare the date and details with another reliable source when the stakes are high.

Takeaway: A deepfake or impersonation may look or sound authentic. Verification depends on the source and the channel, not on how realistic the media appears.

Use It in Real Life: Make an independent check easy

Choose one organization you rely on, such as a utility, employer, school, clinic, insurer, or local government. Save its official web address or phone number from a trusted document now, before you need it under pressure.

If you want AI's help, keep the request limited: “Help me list questions to ask my utility about a rate notice. Do not tell me whether the notice is genuine.” Do not paste account numbers, passwords, or links from a suspicious message.

m7-l1-real-life

The next time a high-stakes claim reaches you, use your saved route to verify it before responding, sharing, paying, or downloading anything.

Key takeaway

m7-l1-key-takeaway

AI can be wrong in a polished way. Treat it as a thinking aid, not a final authority, and verify important claims through a source you independently choose.

Check Your Judgment

Before you choose, pause and use the practical details in the situation. A useful next step is not always the quickest or most polished-sounding one.