𝐀𝐈 𝐂𝐚𝐧 𝐌𝐚𝐧𝐮𝐟𝐚𝐜𝐭𝐮𝐫𝐞 𝐈𝐧𝐟𝐥𝐮𝐞𝐧𝐜𝐞𝐫𝐬.

A man stands on a beach, covered in dirt, shouting that it will not come off.

A woman appears and wipes it clean with a bath mitt.

There is a link below the video. The mitt costs $3.

The woman who posted it has 183,000 followers on Instagram and, according to researchers at Online Risk Labs, probably does not exist.

Fortune reported that 73 of the top 100 creators in Temu’s European partnership ads on Meta showed similar signs. One account had changed its handle 15 times.

The spend behind ads like these, modelled from impressions rather than disclosed, runs to as much as $962 million over 16 months.

Fortune is careful to add that there is no evidence Temu’s ads are fraudulent.

And that’s the point.

Because nobody has to do anything fraudulent for the system to produce this kind of outcome.

Temu optimizes for return on ad spend. Meta is paid for impressions. The account gets a cut of sales.

Everyone is optimizing something except trust.

But at $3, that may not matter much.

The shopper isn’t underwriting the influencer’s judgment. The downside is tiny, and much of the trust can be carried by the platform, payment system and return policy.

Raise the cost of being wrong and the equation changes.

A $3 bath mitt.

A $300 supplement.

A $3,000 investment.

A $30,000 medical decision.

As the cost of being wrong rises, so does the credible stake we demand from whoever asks us to trust them.

That is the variable to watch.

Not realism.

Stake.

A synthetic persona can sell you exactly as much as it can afford to be wrong about.

At $3, the required stake approaches zero. A face with nothing to lose can be enough.

But a face that cannot lose anything cannot cover much downside.

This is why the debate about synthetic influencers risks focusing on the wrong problem.

Stop watching video quality.

Watch for the moment synthetic stake becomes manufacturable.

A brand puts its name behind a person who does not exist.

A guarantee is attached to an invented face.

A platform verifies a synthetic identity.

An insurer underwrites its claims.

That is the upmarket path: real institutions lending artificial people their reputation, balance sheet and liability.

And it will arrive dressed as accountability.

What matters more is not so much:

“Was this person generated by AI?”

But rather:

“Who loses something if I believe them and they’re wrong?”

Trust is not demanded by price.

It is demanded by the cost of being wrong.

Don’t just label the pixels.

Name who pays when they’re wrong.

#ArtificialIntelligence #InfluencerMarketing #Trust #DigitalIdentity #AIRegulation

Contact

bruno.gentil@sherpaconsultingasia.com

© 2026. All rights reserved.