A photograph can be generated, an audio recording can be manipulated and a video can present an event in a misleading way. Students need a checking habit that works even when the content looks convincing. Guessing from appearance alone is a weak starting point.
In Kenya, a claim may reach a learner through a forwarded message, a social feed or a group conversation before anyone has checked its source. The practical response is to pause, identify what is being claimed and look for dependable evidence. You do not need to prove exactly how a piece of content was made before deciding that it should not be shared as fact.
Understand the different kinds of misleading content
AI-generated images, audio and video can imitate a real-world scene or person. A deepfake can make it appear that someone said or did something they did not. Other misleading content may use an authentic image with the wrong date, location or description.
That distinction matters. A real photograph can support a false story, and a synthetic image can be clearly labelled as illustration. Ask what the content is being used to claim rather than treating “AI” as the only possible explanation for deception.
Name the claim before checking it
Write the claim in one sentence: who supposedly did what, where and when? If the message contains no clear source or date, note that gap. Separating the claim from the emotional wording makes it easier to decide what evidence would actually confirm it.
Imagine a video caption claiming that a school has announced a new rule. The useful check is the school’s official communication or a reliable account of the announcement. A dramatic soundtrack and many shares do not establish that the rule exists. Start with the institution that would know.
Find the original source
Ask where the content first appeared. A screenshot of a screenshot can remove dates, context and the account that posted it. A forwarded message may identify no author at all. Search for the original statement or publication rather than relying on the person who most recently shared it.
Look at the source itself. Does it provide evidence, dates and a way to understand the context? An official-looking logo can be copied, so use the organisation’s known website or established channel. If you cannot locate the original, describe the claim as unverified rather than filling in the missing details yourself.
Reverse your first assumption
If your first reaction is “This must be real”, ask what would change your mind. Could it be a demonstration, an old event, a joke or an edited extract? If your first reaction is “This must be AI”, ask whether an original recording or trustworthy report might establish that it is genuine.
This habit reduces the influence of what you already want to believe. The question is not whether a claim feels plausible or fits your opinion. It is what evidence supports this particular account and whether that evidence is being represented accurately.
Notice emotional pressure
Be cautious when a message demands immediate forwarding, says everyone is hiding the truth or uses fear and outrage to discourage checking. Emotional language can accompany real events too, but pressure to act quickly is a good reason to pause.
Ask what the message wants you to do: share, accuse, send money, reveal information or attack someone. A content check and an action check belong together. Even before accuracy is settled, you can refuse to share private material, bully a person or act on an unverified instruction.
Compare independent, reliable sources
Look for multiple dependable sources that have investigated the claim independently. Several accounts copying the same message do not create independent confirmation. For an institutional announcement, a direct statement can be more useful than a collection of reposts.
Read beyond headlines. Confirm the date, location and exact claim. A report about a similar event elsewhere does not verify the message in front of you. If sources disagree, identify the disagreement and ask a teacher or adult to help decide what further evidence is needed.
Use image and video checks as clues
Reverse-image searches can sometimes locate earlier appearances or different context for an image. Key frames from a video may help trace its origin. Search results are leads to investigate, not a guarantee that the content is genuine or fake.
Visual oddities can also prompt questions, but they are not a complete method. Some generated media looks convincing; some genuine media looks strange because of lighting, compression or camera effects. Do not build your whole decision on fingers, shadows or whether a face “looks wrong”. Source and context remain essential.
Be careful with voices and familiar people
A voice that sounds like a familiar person is not enough to justify an urgent request. If a recording or call asks for sensitive information or immediate action, verify through a separate trusted channel. Do not use contact details supplied only inside the suspicious message.
For students, this means involving a trusted adult rather than trying to investigate a worrying request alone. The same habit protects against impersonation more generally, whether AI was used or not. Keep passwords and personal information private while checking.
Protect people while reporting harm
An altered image of a classmate or a fabricated recording can cause real harm. Do not forward it to friends for entertainment or circulate it widely while claiming to expose it. Use the school’s reporting route and share only what the responsible adult needs.
Avoid accusing someone of making a deepfake without evidence. Explain what you received, why it concerns you and where it came from if known. Respectful reporting keeps the focus on protecting the affected person and resolving the concern.
A five-step student routine
- Pause before believing, sharing or acting.
- State the claim clearly, including any missing date or source.
- Look for the original statement or context.
- Compare appropriate, independent evidence.
- Ask a trusted adult when the claim is sensitive or remains uncertain.
Practise with adult-selected fictional material. One example can be a generated scene labelled as illustration; another can be an old photograph attached to a new caption. Ask students to explain their process and what they still do not know. Avoid exercises using embarrassing images of real pupils.
Common questions
Can an AI detector settle the question?
Treat automated outputs as clues requiring review rather than final proof. Your decision should consider source, context and corroborating evidence.
What if I still cannot tell?
Do not share the content as established fact. Say it is unverified and ask for help where appropriate. Uncertainty is a responsible conclusion when evidence is missing.
How can schools teach this regularly?
AI Ready connects verification with privacy and responsible use. Link these habits with student homework research so checking evidence becomes part of ordinary learning.