If a family member shares a political image, video, or claim that may be AI-generated or misleading, check it before correcting them. Then ask where it came from, offer to verify it together, and focus on the evidence rather than their motives. Curiosity makes a more useful starting point than accusation—and if the conversation stops being constructive, pause it.
How to start the conversation
A person can share false material while believing it is true. PEN America distinguishes misinformation shared without knowing it is false from disinformation shared with intent to deceive; a relative’s post alone does not establish their intent. See PEN America’s guidance on talking with friends and family.
As an Amazon Associate I earn from qualifying purchases.
Ask before you correct
Begin with a question about the item, not a judgment about the person. The Indiana Department of Homeland Security puts it simply: “Start from a place of curiosity, not criticism.” You might ask where they saw the video or what made it seem convincing. Or try: “That is interesting. I saw something different from …” and “Can I show you what I found about that?”
Free tools Windows power users keep installed
One-click scans. No signup required.
These questions invite a conversation without pretending that the claim is established. Avoid calling someone gullible, accusing them of lying, or trying to win the exchange. Indiana DHS offers more examples in its misinformation and disinformation guidance.
#1 Best Overall
Make checking it a shared task
Offer to look into the claim together. Keep the discussion on the specific image, clip, or assertion, and bring one or two reliable sources rather than an intimidating pile of links. If you find a correction, explain what the source establishes and what it does not. A correction can be clear without implying that the person who shared the post acted in bad faith.
Verify the item before you challenge it
Suspecting that political content is false or AI-generated is not the same as confirming it. Check the claim itself before raising it with a relative; a visual impression or automated detection result is not enough.
- Trace the source. Look for the earliest available post, account, outlet, or document, and see whether it points to an identifiable source. A repost without context is a weaker basis for a claim than an original document or attributable statement.
- Check date and context. An old image or statement can be recirculated as if it were new. Look for when and where the material first appeared, if that can be established, and whether the surrounding context changes its meaning.
- Seek independent corroboration. Check credible reporting and an independent fact-checker, such as AP Fact Check or PolitiFact. For voting dates, rules, locations, or procedures, use the relevant official election authority rather than a social post, chatbot, or AI-generated search summary.
- Use visual irregularities as leads, not verdicts. Unusual shadows, distorted details, mismatched lip movements, or stiff expressions can justify further checking. They do not prove that a clip is fabricated; realistic AI-generated media may not show obvious errors.
- Interpret detection and provenance cautiously. A detector can be wrong, and provenance information may be missing, invalid, or incomplete. Neither a tool’s score nor the absence of a provenance label settles authenticity. Use such information as context alongside the source, date, and corroboration.
The Brennan Center for Justice cautions: “In general, voters should not depend on deepfake detection tools to verify or disprove the authenticity of political content because the tools have limited accuracy.” Its election-information guidance also explains why provenance is only one part of verification.
Recommended Free Tools
Decide whether to respond privately or publicly
The right setting depends on what you are trying to accomplish. A private message can let a relative reconsider without being embarrassed in front of others. A public correction may help bystanders when a post is already spreading, but replying can also increase its visibility.
Rank #3
| Approach | When it may fit | Main trade-off |
|---|---|---|
| Private message | You want to preserve the relationship and the post has limited reach. | It may not reach people who already saw the original claim. |
| Public correction | The claim is spreading and a correction could help people who encounter it. | Your reply may draw more attention to the original post. |
Before replying publicly, consider the post’s reach and whether engagement is likely to amplify it. PEN America discusses these trade-offs in its conversation guidance.
Know what a correction can—and cannot—do
Media literacy and corrections may help people recognize misleading material, but the evidence does not support promising that one conversation will change someone’s mind or that any one approach works best for everyone. A UK Government Office for Science evidence note published on 27 May 2025, based on an expert roundtable held in March 2024, describes media literacy as useful while noting limits: a cited analysis found that belief corrections did not persist at follow-up, and the note does not establish a universally best intervention. A preregistered study of visual media likewise reported inconsistent earlier findings about whether specific identification tips outperform general advice; its authors said firm conclusions about those tips could not be drawn.
Caltech’s 2026 account of a preregistered US panel experiment reports that researchers tested AI-generated prebunking against five common myths about the 2024 US election among more than 4,000 registered voters. That is evidence about that particular intervention and study context, not proof that a family discussion will have the same effect. Similarly, Brennan Center analysis of tested systems found that models could generate misleading election images and video, often failed to identify AI-generated content, and had uneven safeguards; it also reports cases where systems pushed back against longstanding false election tropes. AI systems are neither reliably truthful nor uniformly deceptive.
When to pause
If the exchange becomes defensive or stops producing useful discussion, step back. You can say you would rather return to it later, or leave a reliable resource for the person to review. Pressing on after the conversation has become a contest is unlikely to improve the quality of the evidence either of you considers.
Quick Recap
Best Value
Further guidance
- PEN America: How to talk to friends and family who share misinformation includes advice on conversation, checking images and sources, and handling breaking news.
- Indiana Department of Homeland Security: Misinformation and disinformation offers practical conversation prompts and verification suggestions.
- Brennan Center for Justice: How to verify election information discusses election sources, deepfake detectors, and provenance.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
