When Seeing Is No Longer Believing: How to Navigate AI on Social Media
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Integer nec odio. Praesent libero. Sed cursus ante dapibus diam. Sed nisi. Nulla quis sem at nibh elementum imperdiet.
Two days ago, I shared something on Instagram that left many of us unsettled.
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Integer nec odio. Praesent libero. Sed cursus ante dapibus diam. Sed nisi. Nulla quis sem at nibh elementum imperdiet.
Category
Lifestyle
Date
04/10/2026
Length
5 min read
Share
Two days ago, I shared something on Instagram that left many of us unsettled.
An account had taken videos made by real women — including creators I know and follow
— and used AI to place a different face over their bodies. The account then built an entirely
fictional woman around those videos: a name, a personality, a lifestyle and a sense of trust.
She appeared to be a real person. She was not.
That fictional identity was then used to sell real products, including supplements and
health-related advice.
The response to my post was overwhelming. Many people were shocked. Others said they
had already noticed more and more AI-generated people online. Some asked a question
that is becoming harder to answer every day:
How are we supposed to know what is real anymore?
I do not think the answer is to become suspicious of every image, video or person we see
online. But we do need a new habit: pause before we trust, buy, share or act.
AI is not only about funny images
When people talk about AI on social media, they often mean harmless things: an edited
background, a picture made for fun or a filter that changes someone’s face.
But the same technology can also be used to copy someone’s appearance, imitate a voice,
create a fake expert or make a person appear to say or do something that never happened.
It can be used to manufacture a whole identity — and then use that identity to sell
products, gather followers or influence people.
That is what makes the example I shared especially disturbing. It was not simply an
artificial image. It appeared to use the work and bodies of real women to create credibility
for someone who did not exist.
The issue is therefore not just, “Was AI used?” The more important questions are:
Whose image, voice or work was used?
Is the person presented as real or fictional?
What does the account want me to believe or buy?
Could I be harmed if this is not true?
AI can be creative. It can also be deceptive. Those two things can exist at the same time.
A label helps, but no label is not proof
You may see labels such as “AI info” or “AI-generated profile” on Instagram. Meta says
these labels can appear when it detects signals from AI tools or when someone discloses
that AI was used. Instagram also has an AI-generated profile label for accounts that
regularly feature AI-generated people. (3)(4)
These labels are useful, but they are not a magic truth detector.
A label may tell you that AI was used. It does not automatically tell you whether the claim
accompanying the content is true, whether the account has permission to use another
person’s likeness or whether a product is safe.
The reverse is also important: the absence of an AI label does not prove that content is
real. Detection systems are not perfect, labels may be missing, and a piece of content can
be copied, re-edited or uploaded in a way that removes information about its origin. Meta
itself explains that its labels depend partly on industry signals and user disclosure.(4)
So treat a label as one piece of information, not as a final verdict.
Five questions to ask before you trust a post
You do not need to become a technology expert. You need a simple checking routine.
1. Who is behind this account?
Look beyond the profile picture and follower count.
Does the account have a history? Are there older posts that show a consistent person, place
and life? Is there a real website, a clear contact option or information about who runs the
account? Do other independent sources refer to this person or organisation?
Be careful with accounts that appeared recently, post only highly polished content or have
a personality that feels strangely perfect. A large audience is not proof of a real identity.
Followers can be bought, copied or built around an artificial character.
A blue tick is not proof that every claim is true either. Verification can tell you something
about an account, but it cannot verify every video, recommendation or health statement
that account publishes.
2. What exactly am I looking at?
Slow the video down if something feels strange. Look at the mouth, teeth, eyes, hairline,
hands and the edges around the face. Does the lip movement match the words? Does the
expression move naturally? Does the lighting on the face match the rest of the image? Do
reflections, shadows and jewellery remain consistent?
These visual clues can still be helpful, but they are becoming less reliable as the technology
improves. Mediawijsheid.nl recommends looking at unnatural mouth movements, missing
or unusual blinking, blurred pixels and a mismatch between the head and the rest of the
body. (2)
The most important warning sign is not always a visual glitch. Sometimes the content looks
completely convincing. The warning sign may instead be the combination of a perfectlooking
person, a strong emotional story and a product that you are encouraged to buy
immediately.
3. Can I find the original?
If you recognise a creator, search for the original account. Look at older posts. Compare the
voice, body, background, clothes, editing style and upload dates.
You can also take a screenshot and use a reverse-image search, or search distinctive words
from the caption. Try to find out where the image or video appeared first and whether the
current account has changed the context.
This matters because a real video can be used to support a false story. The video itself may
be genuine, while the face, voice, caption or product claim has been added later.
4. Is the claim supported elsewhere?
A serious claim should not depend on one anonymous account or one emotional video.
For news, politics or public events, check established news organisations and primary
sources. For health, medication, supplements or medical advice, check a doctor,
pharmacist, public health service or another qualified source. The World Health
Organization emphasises the importance of making credible, science-based health
information easier to find online. (6)
If an account says, “Doctors do not want you to know this,” “This one product fixes
everything” or “Buy now before it is gone,” stop. Urgency and secrecy are common ways
to push people past their own doubts.
5. What is this post asking me to do?
Ask yourself what happens if you believe it.
Are you being asked to buy something, send money, share personal information, click an
unfamiliar link, vote, become angry at a group or forward the post quickly?
The higher the stakes, the higher your standard of proof should be. A funny picture does not
need the same investigation as a health recommendation, a financial offer, a political claim
or a message that could damage someone’s reputation.
When can you assume something is not true?
The honest answer is: you often cannot know with absolute certainty from the post
alone.
Instead of asking, “Can I prove this is fake?” ask:
“Is there enough trustworthy evidence for me to rely on this?”
If the answer is no, treat it as unverified. Do not buy because of it. Do not share it as fact.
Do not repeat a serious accusation. You can say, “I have not been able to verify this,”
without claiming that you know it is false.
There are, however, situations in which you should act as if a claim is unsafe to rely on:
when the account is anonymous or deceptive, when the content uses someone else’s
identity without explanation, when there is no independent evidence, when the post
creates pressure to act immediately or when the claim involves health, money, safety or
politics.
Being careful is not the same as being cynical. It is a way of matching your level of trust to
the evidence available.
How to warn each other without spreading the content
further
When you find a suspicious account, your first instinct may be to repost it so everyone can
see it. That can unintentionally help the account: more views, more comments and more
reach.
A better approach is:
1. Save evidence privately. Take screenshots of the profile, posts, captions, links and
product pages. Record the username and URL. Content can disappear after a report.
2. Contact the real creator privately. If someone’s face, body, voice or work has been
used, send them the evidence directly. Let them decide how they want to respond.
3. Report the account and the specific content. Instagram says impersonation violates
its Community Guidelines and provides a reporting route for accounts pretending to be
someone else.(5)
4. Warn people without linking to the suspicious account if possible. Describe the
pattern, explain what you checked and point people towards the real creator or an
official source. Avoid repeating the sales link.
5. Do not contact or threaten the people behind it. Reporting, documenting and
supporting the people affected is more useful than sending angry messages that may increase the account’s visibility.
6. If money, identity theft or serious harm is involved, get help. Contact your bank, the
relevant platform and the appropriate authorities. For medical concerns, speak to a
healthcare professional rather than relying on comments under the post.
The goal is not to create a bigger spectacle. The goal is to reduce harm.
What social platforms and creators should do
The responsibility cannot sit entirely with individual users. Platforms need better detection,
clearer labels, faster reporting routes and stronger protection for people whose content or
likeness is being misused.
The European Union’s AI Act includes transparency obligations for certain AI systems,
including obligations relating to machine-readable marking and the disclosure of
deepfakes. Article 50 applies from 2 August 2026. (1)
That is an important step, but a law or label does not remove the need for enforcement. Nor does it mean that every piece of content online will suddenly be easy to verify.
Creators can help by clearly stating when an image, voice or person is fictional. If AI is used
to alter a real person, that should not be hidden behind a vague caption. If an account is
entirely fictional, viewers should not have to investigate to discover that fact.
And if you use somebody else’s video, face or voice, you need to think about permission,
context and potential harm — not only what a tool makes technically possible.
A calmer way forward
I understand why people feel frightened by this. When we can no longer trust our eyes
automatically, it can feel as if everything is becoming meaningless.
But the answer is not to stop believing anyone. It is to stop treating appearance as evidence.
We can still enjoy beautiful images. We can still discover new people and ideas. We can still
be open, curious and inspired. We just need to add a pause before trust becomes a
purchase, a share or a judgement about another person.
Look at the source. Check the context. Find the original. Compare independent
information. Be especially careful when health, money, politics or someone’s reputation
is involved.
And when something feels wrong, you do not need to solve the entire future of AI by
yourself. Save the evidence, warn the people affected and report it.
The future of social media may not allow us to believe everything we see. But it can still
allow us to make thoughtful decisions about what we choose to believe, share and support.
A quick checklist to save
Before you trust or share a post, ask:
Source: Who created this, and can I verify who they are?
Original: Where did this image or video first appear?
Evidence: Do independent, trustworthy sources support the claim?
Interest: What does this account want me to buy, believe or do?
Risk: What could happen if it is wrong?
Pause: Have I checked before reacting or sharing?
If you cannot answer those questions, you do not have to decide immediately. Pause is a
perfectly good response.
References
[1] European Commission: Guidelines on transparency obligations for providers and
deployers of certain AI systems
[2] Mediawijsheid.nl: Zien is geloven? Dit moet je weten over deepfake video's
[3] Instagram Help Center: About AI-generated profiles on Instagram
[4] Meta: Our Approach to Labeling AI-Generated Content and Manipulated Media
[5] Instagram Help Center: Impersonation Accounts
[6] World Health Organization: Global principles for identifying credible sources of health
information on social media
[7] AndBloom Instagram post: When seeing is no longer believing
