Fake shops built in minutes, cloned voices of relatives and deepfake videos of trusted local figures. What to watch for this season, and how to protect your money.
The Caribbean AI Risk Management Council (CAIRMC) expects AI-enabled scams to rise across the region this holiday season. With the Caribbean AI Association, it estimates that about 1 in 6 people in 16 Caribbean countries, roughly 5.1 million, will be targeted by an AI-enabled scam before the new year, and that AI-enabled cybercrime could cost the region about US$350 million in 2026. Norton's 2026 holiday report found that one in six US adults has been scammed while holiday shopping online, and blocked fake shopping websites roughly doubled in a year. This guide explains the numbers, the five AI scams to watch for and how to protect yourself.
Part 1 of 7One in six US adults was scammed while holiday shopping
The best recent measure comes from Norton, the security company, which surveyed 1,000 US adults in September 2026. Thirty percent said a scammer had targeted them while they were holiday shopping online, and more than half of those fell for it. Overall, 16 percent of adults were scammed, rising to 25 percent of Gen Z. Of those scammed, 88 percent lost money, US$394 on average.
The survey also shows why scams work. Thirty-five percent of people said they had bought from a website they suspected was a scam because the deal was too good to miss, and 65 percent would buy from a site they had never heard of if the price was right.
“Scammers are counting on holiday shoppers being rushed, distracted and afraid of missing out”
Leyla Bilge, global head of scam research, Norton
Why it matters
No equivalent survey exists for the Caribbean, so these US figures are a guide. But Caribbean shoppers use the same websites, social media platforms and AI tools, and many buy from US and Chinese online stores for Christmas.
The season also brings more money moving between families: remittances, gifts and travel bookings. Each of those is a moment when a convincing fake can pass for the real thing.
Part 2 of 7AI is building fake shops faster than they can be shut down
A fake online shop used to take days to build and often looked amateur. With AI, a scammer can produce polished product photos, reviews and descriptions in minutes.
Norton's parent company, Gen, says the number of fake shop websites it blocked rose from about 260,000 in the 2024 holiday quarter to 526,000 in 2025, and that fake-shop activity was up 129 percent at the start of this season. The fake shops also last: in the United States, one typically stays online for about seven months.
Why it matters
Fake shops advertise in search results and social media feeds, the places most people look for bargains. Forty-seven percent of US holiday scam victims were caught on a social media platform, according to Norton.
Even AI shopping assistants can be fooled. Norton's researchers found an AI assistant recommending a US$279 jacket from a fraudulent store alongside real retailers, with no warning. Sixty-three percent of Americans have used an AI shopping assistant, and 42 percent plan to use a tool like ChatGPT for holiday shopping.
Part 3 of 7CAIRMC and CAIA project US$26.6 billion in reported cybercrime losses for 2026
To estimate where this season is heading, the Caribbean AI Risk Management Council and the Caribbean AI Association fitted a simple trend to the FBI's yearly totals of reported cybercrime losses in the United States: about US$10.3 billion in 2022, US$12.5 billion in 2023, US$16.6 billion in 2024 and US$20.9 billion in 2025. The curved trend fits those four years closely and points to about US$26.6 billion in 2026, a rise of about 27 percent. A more cautious straight-line trend gives US$24.0 billion.
| Measure | Latest actual | 2026 projection | Method |
|---|---|---|---|
| Reported US cybercrime losses | US$20.9B (2025) | US$26.6B, low case US$24.0B | Curved and straight-line trends on FBI totals, 2022 to 2025 |
| Fake shop websites blocked in the holiday quarter | 526,000 (2025) | about 1.05 million | If last year's doubling repeats; scenario only |
The fake-shop figures move faster. Fake shop websites blocked in the holiday quarter doubled between 2024 and 2025. If that pattern repeats, about 1.05 million could be blocked this season. Treat that figure as a scenario only: two years of data are too few to fit a reliable curve.
Why it matters
Both projections point the same way: more money lost, to more convincing scams, this Christmas than last. The US figures are the clearest available signal for the Caribbean, because the region's shoppers use the same platforms and face the same scam operations.
No Caribbean agency publishes yearly fraud-loss totals in a comparable form, so the region's own figures have to be estimated from outside data. The next section does that. CAIRMC recommends that banks, regulators and police begin publishing standard annual figures, so the region can measure the problem it is trying to stop.
Part 4 of 7About 1 in 6 Caribbean people will be targeted by an AI-enabled scam this Christmas
How many people in the region does this reach? CAIRMC and the Caribbean AI Association built a simple estimate for 16 Caribbean countries with about 31 million people. Roughly 22.6 million of them use the internet. If the Caribbean sees the same holiday scam targeting rate as the United States (30 percent), and three in four of those attempts use AI, about 5.1 million people will be targeted this season. That is 1 in 6 people in the region, and about 1 in 4 of those who are online.
The range is wide because the inputs are uncertain. A cautious case (20 percent targeted, half of attempts using AI) gives 1 in 14. A high case, using KnowBe4's finding that 86 percent of phishing attacks are now AI-driven, gives 1 in 5. As a cross-check, Hiya's 2026 survey found that one in four Americans received a deepfake voice call in the past year.
| Step | Input | Value | Source |
|---|---|---|---|
| 1. People in 16 Caribbean countries | Population, 2025 | 31.0 million | World Bank |
| 2. People online | 73% use the internet | 22.6 million | World Bank, latest year |
| 3. Targeted by a holiday shopping scam | 30% this season (low case 20%) | 6.8 million | Norton Holiday Report 2026, US adults |
| 4. Scam attempt uses AI | 75% (low 50%, high 86%) | 5.1 million | CAIRMC and CAIA assumption; KnowBe4 finds 86% of phishing AI-driven |
| 5. Result | Share of all Caribbean people | 1 in 6 | CAIRMC and CAIA estimate |
By country, exposure tracks internet use. In Jamaica, the Dominican Republic and The Bahamas, where about nine in ten people are online, the central estimate is about 1 in 5.
| Country | People online | Likely targeted | Share of population |
|---|---|---|---|
| Haiti | 48% | 1,282,100 | 1 in 9 |
| Dominican Republic | 91% | 2,358,800 | 1 in 5 |
| Jamaica | 90% | 575,200 | 1 in 5 |
| Trinidad and Tobago | 82% | 252,800 | 1 in 5 |
| Guyana | 83% | 156,100 | 1 in 5 |
| Suriname | 87% | 125,800 | 1 in 5 |
| Belize | 80% | 76,100 | 1 in 6 |
| The Bahamas | 92% | 83,900 | 1 in 5 |
| Barbados | 70% | 44,700 | 1 in 6 |
| Saint Lucia | 68% | 27,600 | 1 in 7 |
| Curaçao | 68% | 24,000 | 1 in 7 |
| Grenada | 70% | 18,500 | 1 in 6 |
| Aruba | 97% | 23,800 | 1 in 5 |
| Saint Vincent and the Grenadines | 76% | 17,100 | 1 in 6 |
| Antigua and Barbuda | 73% | 15,400 | 1 in 6 |
| Dominica | 82% | 12,200 | 1 in 5 |
| Caribbean (16 countries) | 73% | 5,094,000 | 1 in 6 |
Why it matters
Applied to money, the same approach suggests the Caribbean could lose about US$470 million a year to reported-equivalent cybercrime in 2026, of which about US$350 million could involve AI (range US$240 million to US$400 million). That is money leaving household savings, small business accounts and remittance flows.
These are estimates built from public data, not official counts. Their purpose is to give banks, regulators and ministries a working figure to plan around until the region measures its own losses.
| Scenario | Share of losses involving AI | AI-enabled losses | All cybercrime losses |
|---|---|---|---|
| Low | 50% | US$240 million | US$470 million |
| Central | 75% | US$350 million | US$470 million |
| High | 86% | US$400 million | US$470 million |
Part 5 of 7Deepfakes have already reached Jamaica and Trinidad and Tobago
The Caribbean is not starting from zero. Criminals have already used AI-generated video and fake articles of well-known local figures to promote fake investment schemes.
| Indicator | Figure | Source |
|---|---|---|
| Identity fraud, Latin America and the Caribbean | +13.3% in a year | Sumsub Identity Fraud Report 2025 to 2026 |
| Deepfake incidents, Latin America and the Caribbean | +255%, 2023 to 2024 | Sumsub regional study |
| Deepfake investor alerts, Trinidad and Tobago | 6 alerts, Oct 2025 to Jun 2026 | Trinidad and Tobago Securities and Exchange Commission |
| Deepfake of businessman Adam Stewart promoting fake returns | Police warning, July 2025 | Jamaica Constabulary Force, via Jamaica Observer |
| Scam centres spreading to Latin America and the Caribbean | Warning, July 2026 | UNODC and the EU Anti-Trafficking Coordinator |
In Jamaica, police warned in July 2025 about a deepfake video of businessman Adam Stewart promoting an investment that promised returns of up to J$1.3 million a month. In Trinidad and Tobago, the Securities and Exchange Commission issued six alerts between October 2025 and June 2026 about fake videos and articles using former leaders, bank chairmen and the current finance minister.
Why it matters
Holiday scams use the same tools. A deepfake video of a popular radio host announcing a Christmas giveaway, or a fake bank chairman promoting a festive investment bonus, needs nothing more than the technology already used in these cases.
CAIRMC's analysis of scam centres also warns that organized scam operations, which the UN says have forced more than 300,000 people into fraud work in Southeast Asia, are spreading towards Latin America and the Caribbean.
Part 6 of 7Five AI scams to watch for this Christmas
Most holiday scams fall into five groups, each made more convincing by AI:
- Fake online shops. Professional-looking stores selling brand-name goods, electronics or toys at prices well below normal, which take your money and send nothing.
- Scam ads on social media. Sponsored posts on Facebook, Instagram and TikTok pointing to those shops, often with AI-made videos of happy customers.
- Deepfake endorsements. Fake videos or articles of trusted local figures promoting giveaways, raffles or investment bonuses.
- Cloned voices of family. A call or voice note that sounds like a relative abroad asking for urgent money. CAIRMC has documented voice-cloning cases hitting Caribbean businesses.
- AI shopping assistants. Chatbots that recommend a product from a fake store because they cannot always tell it apart from a real one.
Why it matters
The common thread is urgency. Every one of these scams works by pushing you to act before you check.
Part 7 of 7What households and businesses should do now
CAIRMC's published guidance for financial institutions, set out in its analysis of deepfake fraud, includes adding AI impersonation to staff fraud training, building fast ways to verify videos with the people likely to be impersonated, and using identity checks designed to catch AI-generated video, not only photos.
For households, the defence is a routine: pause, check, pay safely and report.
Why it matters
Banks, credit unions, money transfer companies and retailers see scams first. If they warn customers before the Christmas peak, and tell them exactly how the bank will and will not contact them, they can stop losses that no police force can recover later.
Eight ways to protect your money this holiday season
- Check the web address. Look for spelling tricks in the shop's address, and search the shop's name with the word scam before you pay.
- Be suspicious of deep discounts. If brand-name goods are 70 percent off on a site you have never heard of, assume it is fake until proven otherwise.
- Pay by card, never by gift card. Credit cards and trusted payment platforms let you dispute a charge. Gift cards, phone credit and transfers to strangers cannot be recovered.
- Agree a family code word. Pick a word only your family knows, so a relative can confirm an urgent call for money is real.
- Call back on a known number. If a bank, courier or relative asks for money, hang up and call them on a number you already have.
- Treat celebrity giveaways as fake. Public figures in Jamaica and Trinidad and Tobago have already been deepfaked. Check their official pages before believing any offer.
- Double-check AI shopping advice. If a chatbot recommends a store, visit the brand's official site or a retailer you know instead.
- Businesses: warn customers now. Tell customers how you will contact them and what you will never ask for, before the Christmas rush.
The holiday scam lesson for the Caribbean
Questions readers ask about AI holiday scams
The 10 sources behind this article
- World Bank, World Development Indicators: population, internet use and GDP per capita (PPP), accessed November 2026
- SecurityBrief, "KnowBe4 finds 86% of phishing attacks now AI-driven," 3 May 2026
- Hiya, "State of the Call 2026: AI deepfake voice calls hit 1 in 4 Americans," Business Wire, March 2026
- FBI Internet Crime Report 2025, summary of losses and complaints, April 2026
- Anvilogic, FBI IC3 2024 report: US$16.6 billion in losses, up 33% on 2023
- TechRadar, FBI IC3 2023 report: losses above US$12.5 billion, up 22% on 2022
- Norton, "Norton Holiday Report 2026: 1 in 6 Americans fell victim to online shopping scams," 5 November 2026
- Caribbean AI Risk Management Council, "The UN just warned scam centres are spreading to the Caribbean," 2026
- Caribbean AI Risk Management Council, "FATF named AI deepfakes a money-laundering risk. Trinidad had already logged six cases," 26 August 2026
- Caribbean AI Risk Management Council, "Your CEO never said that: the AI voice cloning crisis hitting Caribbean business"