Trinidad's Finance Minister Was Deepfaked Twice in 2026. AI Can Stop the Next One.

Twice in 2026, someone in Trinidad and Tobago watched a video of Finance Minister Davendranath Tancoo personally recommending a foreign exchange trading platform. The voice sounded right. The face moved right. The minister had never said any of it.

On 10 April 2026, the Trinidad and Tobago Securities and Exchange Commission (TTSEC) issued a public alert about a fake AI-generated video circulating on Facebook that showed Minister Tancoo promoting an investment opportunity. Two months later, on 15 June 2026, the Ministry of Finance issued a second advisory about a fresh version of the same scheme, this time naming specific fraudulent entities, a supposed trading agent called "Nadia Sanchez" and a platform called "Bullfield Holdings." The ministry called the video "completely fake, unauthorised, and malicious." It urged anyone who received it to report it to the Trinidad and Tobago Police Service, Facebook, and Instagram.

That the same scam returned twice, using the same face, in the same country, within a single quarter, is the real story here. A deepfake is not a one-off prank. It is a reusable weapon, and once a fraud network has a convincing synthetic version of a trusted public figure, nothing stops them from running it again, and again, until someone builds a system that catches it faster than they can post it.

The Anatomy of a Deepfake Financial Scam

The mechanics are straightforward, which is exactly why this problem scales so fast. Fraudsters feed publicly available footage of a real, recognisable person, a minister's parliamentary appearances, a bank executive's television interview, a celebrity's social media clips, into an AI video generation model. Given enough source material, today's models can reproduce a person's face, expressions, and voice closely enough to fabricate new footage of them saying things they never said.

In the Trinidad case, the fabricated footage placed a familiar, trusted face in front of an unfamiliar, fraudulent financial product. That single substitution does most of the persuasive work. A stranger promising extraordinary forex returns gets ignored. The country's own Finance Minister, appearing to say the same thing, gets shared, forwarded, and believed, at least by enough people to make the campaign profitable.

The video is then pushed through paid social media advertising and organic sharing, designed to reach as many people as possible before a regulator, bank, or platform can intervene. TTSEC's April alert and the Ministry of Finance's June advisory both arrived after the videos were already circulating, which is the structural weakness at the centre of this entire problem: verification happens after the harm, not before it.

This Is Not an Isolated Incident

The deepfaked Finance Minister sits inside a wider surge of AI-enabled financial crime hitting Trinidad and Tobago and the wider Caribbean in 2026.

Speaking at CANTO Connect 2026, the telecommunications industry's 42nd annual general meeting held at the Hyatt Regency in Port of Spain from 1 to 3 February, Shiva Bisseesar, managing director of Pinaka Consulting, told the gathering that business email compromise, where fraudsters impersonate vendors or executives to trick employees into authorising payments, is "very prevalent" in Trinidad and Tobago. His warning about the sophistication of modern fraud networks was blunt: "These guys are so good at trying to get you to bypass your existing workflows."

Trinidad and Tobago's Financial Intelligence Unit has separately flagged a rise in romance scams targeting citizens through dating apps and social media, often building months-long relationships before requesting money. Financial sextortion, where criminals coerce victims into sending compromising images and then extort them, is described as an emerging threat concentrated among younger users. Ransomware operations, meanwhile, increasingly run with the internal structure of a legitimate business, complete with specialised divisions for intrusion, negotiation, and payment laundering.

None of this is unique to Trinidad and Tobago. The Financial Action Task Force (FATF), the global body that sets anti-money laundering standards, published a paper on 24 February 2026 examining the scale of cyber-enabled fraud worldwide. It found that 90 percent of the 156 jurisdictions it assessed now identify fraud as a major money laundering risk. As one marker of how fast the problem is moving, the paper cited a 61 percent increase in cyber-enabled scam cases in Singapore over two years. In the United Kingdom, fraud now accounts for more than 40 percent of all recorded crime, and an estimated 80 percent of that fraud is cyber-enabled.

A useful illustration of how far the video-impersonation tactic can go comes from Hong Kong, where an accounts controller at a multinational firm was deceived on a video conference call by deepfaked colleagues, including someone who appeared to be the company's chief financial officer, and authorised transfers totalling roughly USD 25 million. Trinidad's scheme was smaller and cruder by comparison, a single pre-recorded clip rather than a live interactive deepfake, but the direction of travel is the same one that produced the Hong Kong case: synthetic video is becoming good enough, and cheap enough, to impersonate anyone with enough public footage available online. Every Caribbean minister, bank chief executive, and well-known businessperson has that footage available online.

How AI Can Stop What AI Makes Possible

The tools that make deepfake fraud possible are the same category of tools that can stop it. This is not a hypothetical future capability. Deepfake detection, content authentication, and fraud-pattern analysis are deployed today by major banks, social platforms, and news organisations elsewhere in the world. The Caribbean's task is to bring them home and apply them to exactly the kind of incident Trinidad and Tobago experienced twice in 2026.

1. Deepfake Detection at the Point of Upload

Modern deepfake detection models are trained on enormous libraries of both authentic and AI-generated video, learning to spot the artefacts that generation tools still leave behind: unnatural blinking patterns, inconsistent lighting across the face, audio that does not quite match lip movement at a frame level, and spectral irregularities in synthetic voice audio that a human ear misses but a trained model catches instantly.

Deployed as a scanning layer across social media platforms and paired with automated monitoring for a public figure's name and likeness, this kind of detection can flag an impersonation video within minutes of it going live, rather than waiting for a member of the public to notice, report it, and wait for TTSEC or the Ministry of Finance to issue a press release days later. Speed is the entire point. Every hour a fraudulent video stays live is an hour more people see it and act on it.

2. Content Provenance for Government and Bank Communications

The Coalition for Content Provenance and Authenticity, known as C2PA, is an open technical standard, backed by major technology and media companies, that attaches a signed, tamper-evident record to genuine video and images at the moment of creation. Every official video a ministry, central bank, or securities regulator releases could carry this credential, recording exactly when, where, and by whom it was produced.

Once that standard is in place, verification becomes trivial. A citizen, a journalist, or a platform's own automated system can check whether a video carries a valid, unaltered credential. A video purporting to show Minister Tancoo endorsing a forex platform, carrying no credential at all, is immediately and mechanically identifiable as unverified, without anyone needing to analyse the footage frame by frame. This flips the burden of proof from "prove this is fake" to "this was never proven real," which is a far faster and more defensible standard for a regulator to act on.

3. AI-Powered Transaction Anomaly Detection

Even a deepfake video that slips past detection still needs a bank account, a wallet, or a payment rail to extract money from victims. AI-powered anomaly detection systems, already used by major international banks, monitor transaction patterns in real time and flag transfers to accounts, platforms, or beneficiaries already linked to reported scams, unusual transaction velocity, or behaviour inconsistent with a customer's history.

For Trinidad and Tobago's banking sector and Financial Intelligence Unit, this means that even when the Bullfield Holdings video reaches a potential victim, a well-tuned transaction monitoring system can intercept the actual transfer before funds leave the country, or flag it for review before it clears. This layer matters because it does not depend on catching the video at all. It catches the money.

4. Biometric Liveness Verification

The Hong Kong case succeeded because a live video call, one of the most trusted verification channels in modern business, was compromised by real-time deepfake technology. Biometric liveness verification, which checks for the micro-signals of a genuine live human presence rather than a recorded or synthetically generated one, closes that specific gap. Financial institutions requiring high-value transaction authorisation can add a liveness check as a standard step, making it substantially harder for a fraud network to substitute a synthetic executive or client into the approval chain.

What Trinidad and Tobago Is Already Doing

To its credit, the government has not been standing still. Around 10 July 2026, Trinidad and Tobago formally became the 20th member of LAC4, the Latin America and Caribbean Cyber Competence Centre, a regional cybersecurity body supported by the European Union. The agreement was signed by Minister of Homeland Security Roger Alexander, with the Ministry of Homeland Security serving as the country's host institution for the partnership.

LAC4 Director Liina Areng welcomed the addition, saying the centre was "pleased to welcome Trinidad and Tobago," adding that bringing together diverse experiences across the region "enhances cyber resilience." EU Ambassador Cécile Tassin described the accession as "more than a milestone," calling it "a clear commitment to stronger cyber resilience, deeper regional cooperation." Minister Alexander said the government was "confident that this partnership will contribute meaningfully" to national cybersecurity capacity.

Membership gives Trinidad and Tobago structured access to shared cyber threat intelligence, joint training, and policy development support from other member states across Latin America and the Caribbean. That regional intelligence-sharing layer matters directly for deepfake fraud, since these campaigns rarely originate from, or stay confined to, a single jurisdiction.

On the legislative side, an inter-ministerial committee has been examining how to update Trinidad and Tobago's law to address AI-enabled cybersecurity threats, with the Attorney General's Office identified as the natural lead for amending legislation to criminalise these offences. The scale of what is on the table has already reached the country's civic education pipeline. During the 2026 Cohort of the Youth Electoral Success (YES!) Leadership Programme, young parliamentarians in a mock sitting debated a motion to introduce criminal penalties for using AI to commit fraud or create harmful deepfakes, alongside fines for companies that fail to protect customers' personal data. It was a training exercise, not a passed law, but it is a useful signal that the issue has moved from technical circles into the country's next generation of policymakers.

What the Rest of the Region Should Do

Trinidad and Tobago's experience is a preview, not an outlier. Every Caribbean territory with a national bank, a securities regulator, and public officials who appear on camera carries the same exposure. Three actions would move the region from reactive alerts to proactive defence.

First, every Caribbean central bank and securities regulator, following TTSEC's lead, should stand up a dedicated deepfake and impersonation monitoring function, whether built in-house or contracted from a specialist vendor, that scans continuously for synthetic media using a minister's, governor's, or institution's name and likeness. Waiting for public reports, as both Trinidad incidents show, means the video has already been circulating for hours or days by the time an official advisory goes out.

Second, CARICOM member states should coordinate a regional content provenance standard for government and financial sector communications, so a citizen in Kingston, Bridgetown, or Port of Spain can apply the same verification check to an official video regardless of which government produced it. A fragmented, country-by-country approach to authentication makes it easy for fraud networks to simply target the jurisdictions without a standard in place.

Third, Caribbean banks and financial intelligence units should treat transaction anomaly detection as a frontline defence, not a back-office compliance function. The Bullfield Holdings scheme, and every scheme like it, ultimately needs a functioning payment channel to succeed. A system that reliably interrupts that channel makes the video itself far less profitable to produce in the first place, which is the kind of economic disincentive that changes fraud networks' behaviour faster than any single public alert.

The Trust at Stake

What is genuinely at risk here goes beyond the money any single victim might lose to a fake forex platform. Every deepfake of a sitting Finance Minister, left unaddressed, erodes the basic assumption that a video of a public official saying something is evidence that the official actually said it. That assumption underpins how citizens receive emergency instructions, verify government announcements, and trust the institutions managing their money.

Trinidad and Tobago's TTSEC and Ministry of Finance responded to both 2026 incidents quickly and clearly, and that transparency matters. But a public advisory issued after a video has already spread is damage control, not prevention. The next phase of this fight has to be built on AI systems that catch the fake before the public ever sees it, and that catch the money before it ever leaves a Caribbean bank account. The technology to build that system exists now. The region's task is to decide to build it before the next deepfake, not after.

Frequently Asked Questions

What happened with the deepfake video of Trinidad and Tobago's Finance Minister?

In April and again in June 2026, an AI-generated deepfake video of Finance Minister Davendranath Tancoo circulated on Facebook. The video used manipulated audio and video to mimic his voice and likeness, falsely showing him endorsing a foreign exchange trading and investment scheme linked to the names "Nadia Sanchez" and "Bullfield Holdings." TTSEC issued a public alert on 10 April 2026, and the Ministry of Finance issued its own advisory on 15 June 2026, confirming the video was fake and unauthorised and urging the public to report it to the Trinidad and Tobago Police Service, Facebook, and Instagram.

How do AI deepfake financial scams actually work?

Scammers feed publicly available footage and audio of a real person, often a minister, bank executive, or celebrity, into an AI video generation model. The model learns the person's face, mannerisms, and voice well enough to produce new video in which that person appears to say things they never said. In the Trinidad case, fraudsters made it look as though the Finance Minister was personally endorsing a private trading platform, then pushed the fabricated video through social media to reach as many people as possible before a bank, regulator, or platform could catch it.

How can AI detect deepfake videos before they spread?

AI deepfake detection tools analyse video and audio for the artefacts that generation models leave behind: inconsistent blinking, unnatural facial lighting, mismatched lip-sync timing, and audio spectral signatures that do not match a genuine human voice. Detection models trained on large libraries of real and synthetic media can flag a deepfake in seconds. Paired with automated monitoring of social platforms for a public figure's name and likeness, this lets a ministry or bank catch an impersonation video within minutes of it going live, rather than waiting for public reports days later.

What is content provenance and how does it stop deepfake fraud?

Content provenance uses a cryptographic standard, such as the C2PA framework, to attach a tamper-evident record to genuine video and images at the moment they are recorded. Every official video released by a ministry or bank would carry this signed credential, showing when, where, and by whom it was created. A viewer or platform can then check whether a video carries a valid credential. A video claiming to show the Finance Minister that carries no valid credential is immediately identifiable as unverified, without needing to analyse the footage itself.

Is Trinidad and Tobago the only Caribbean country facing AI-enabled fraud?

No. FATF's February 2026 paper on cyber-enabled fraud found that 90 percent of the 156 jurisdictions it assessed now identify fraud as a major money laundering risk, and cited a 61 percent rise in cyber-enabled scam cases in Singapore over two years. In Trinidad and Tobago, business email compromise is described as very prevalent, the Financial Intelligence Unit has flagged a rise in romance scams, and financial sextortion targeting younger users is an emerging concern. Every Caribbean territory with a banking sector and public officials with a media presence carries the same exposure.

What is LAC4 and how does it help Trinidad and Tobago's cybersecurity?

LAC4, the Latin America and Caribbean Cyber Competence Centre, is a regional body that helps member states build cybersecurity capacity through training, joint incident response, and policy development, supported by the European Union. Trinidad and Tobago formally joined as its 20th member around 10 July 2026, with the agreement signed by Minister of Homeland Security Roger Alexander. Membership gives Trinidad and Tobago access to shared cyber threat intelligence from other member states, which is directly relevant to detecting AI-enabled fraud campaigns that rarely respect national borders.

What should Caribbean banks and regulators do about AI-generated scam videos?

Caribbean banks and regulators should deploy three layers of defence. First, automated deepfake and impersonation monitoring that scans social platforms continuously, rather than relying on public reports. Second, AI-powered transaction anomaly detection that flags transfers to accounts already linked to reported scams. Third, a public content provenance standard for official communications, so citizens and platforms can verify a video's authenticity in seconds rather than waiting for a press release days after the damage is done.

Can ordinary Caribbean citizens verify if a video is a deepfake themselves?

Citizens can apply some manual checks, watching for unnatural blinking, mismatched lip movement, and audio that sounds slightly robotic. But manual checks are unreliable against modern generation models. The more dependable path is verification: checking whether a video was posted from the official, verified account of the ministry, bank, or public figure in question, and treating any investment endorsement delivered through a shared video, rather than an official statement, as suspicious by default. Regulators such as TTSEC publish reporting channels specifically for this reason, and using them quickly helps AI monitoring systems flag the same content faster for everyone else.