What Happens When Your Likeness Gets Used For A Deepfake Scam?

Early this year, a Facebook ad used a video of Filipino Co-Chairman and CEO of Doubledragon Corporation, Edgar "Injap" Sia II, to promote an investment opportunity guaranteeing a weekly payout of ₱96,500 (around $1,500 equivalent). The link went to a fake inquirer.net website, one of the most influential newspapers in the Philippines. If instead of Sia it was you in the video, and you woke up with hundreds of Facebook notifications and DMs asking if you authorized this, your stomach sinks to the floor.

These deepfakes are not an unusual one-off targeted to one individual. Rather, scammers use a recycled formula used to target a vulnerable public desiring a specific outcome. Doesn't matter if the victim is a business magnate used to promote a crypto scam, or a surgeon recommending a medical product. Corporate communications will get the videos removed. But the videos always come back, this time under a different figure.

Whether the residual reputational damage is permanent or just a minor bump in the road depends largely on what steps you took before the deepfake incident.

The video lives for days. What the machine answers lives for years.

Sia's Communications team handled the ad just fine. Meta was notified and promptly removed it. SEC was informed. PressOne, an independent news and information website, fact checked and verified the ad was false. But outside of his team's control a cascade of events happened.

Individuals that were duped by the advertisement and lost money complained online, linking his name to the scam. Those threads get indexed and live forever.

Search queries mapping "Sia" and "scam" started to show up in suggested queries, in news archives, and in AI answers where the assistant either searches the live web or draws on whatever was already public when it was trained.

If six months from now a fund manager, or journalist, or counterparty searches his name in ChatGPT, Google, Perplexity, etc., and asks if he was involved in an investment scam, the answer comes back in seconds.

That answer arrives pre-loaded with bias based on everything that was published up to that point, in what order, and by whom.

The removal is not the outcome. The machine's answer is.

The record before the attack determines the outcome.

For a taipan like Edgar Sia II, the scam is but a minor footnote. The man has over two decades of respected news coverage. He is listed on Forbes' Richest list, has a profile about him on Tatler Asia, co-founded a brokerage, and is the chairman of a listed company. His public record is so dense the deepfake scam bounces off it like a rubber ball.

But what about the founder that's a level or two below Sia? This individual may have a LinkedIn, a company bio, maybe a few podcast appearances or press releases. The deepfake scam is no longer just a footnote. Instead, it easily takes up a third of page one on Google and at least a full paragraph or more in AI queries about the founder.

In an effort to be impartial, the machine takes the average of what is publicly available on the internet to generate a response. For a well-defined public figure like Sia, the answer resolves itself easily. But for a founder with a thin public profile, the scam becomes part of the answer whether it's factually true or not.

An agency can take down a link, a post, a video, no problem. What they can't sell you is your reputation after the take-down incident is handled. The foundation must have already been laid well before the deepfake ever surfaced. In other words, the deepfake does not create a reputation problem so much as expose one.

What a correct response looks like.

The public record is full of deepfake incidents and financial scams. All of them track the same way when handled correctly.

Preserve evidence before killing the post.

Once Meta, YouTube, TikTok, etc. removes the post, your evidence is gone along with it. Before anything else, document everything.

• Meta Ad Library information if available
• Ad URL and Facebook page/account
• Download the video or screen record it in full
• URLs of the fake landing pages or news sites
• Domain information if available
• Any phone numbers, WhatsApp/Telegram accounts, payment instructions, or bank/crypto account numbers
• Engagement metrics and comments

Establish one canonical version of reality.

This should ideally come from an authoritative page that you own rather than social media. LLMs can't track posts on social media like they do web pages on a website. Your statement might get seen by your audience on X, Facebook, or LinkedIn, but then largely ignored by Google and AI models.

The statement needs to be the canonical source that all other responses point back to. If a news outlet reports on the scam, they should have the exact page on your website to reference that provides the full account of the incident. No dilution.

Also, and this is easy to mess up, the page must have every form of your name. So for Edgar it would be "Edgar Sia." "Injap Sia." "Edgar Sia II." Each name variant provides a different search query, and the one that you don't have listed is the one the scam threads rank for above your canonical version. Try to anticipate the search language people will use to query the scam incident and include those phrases in the page.

Third-party corroboration.

The PressOne fact-check page provides more armor to Edgar Sia's name than anything he could publish himself. Your record is the one that gets read by news outlets. Their record is the one that gets trusted by LLMs and the broader public. For a prominent figure like Sia, every news outlet in the Philippines gave him free coverage of the incident. For a lesser-known founder, obtaining the vital third-party corroboration takes time and diligence.

The work you do proactively to inform the public record on your name, as well as the quality of work done within the first 72 hour window after the deepfake surfaces, completely determine how AI search will render you when you are not in the room to explain yourself.

Imagine six months down the line an investor, journalist, or prospective business partner asks their favorite AI assistant:

"Was Edgar Sia involved in an investment scam?"

The response goes 1 of 2 ways.

"There have been allegations involving Edgar Sia…"

or:

"No. Scammers used a manipulated video of Sia to promote a fraudulent investment scheme; independent fact-checkers confirmed the endorsement was fake."

That is reputation architecture.