Oil to AI: A Guyana AI-native Economy
Oil revenue is letting Guyana wire artificial intelligence into policing, health and identity faster than any other Caribbean state. The country has passed a data-protection law but no dedicated AI rules, and one of the newspapers that raised the alarm has already closed.
Guyana is deploying artificial intelligence across policing, healthcare and identity faster than any other Caribbean country, funded by its oil boom. Government figures show G$3.6 billion (about US$17 million) spent on AI-enabled video surveillance in the first half of 2026 alone, with over 20,000 cameras and nationwide crime-prediction tools planned by 2030. By mid-2026 the country had passed only the Data Protection Act 2023, with no dedicated AI law before Parliament. The systems are arriving faster than the rules, which leaves Guyana the first country in the region to face a question the others have been able to postpone: when an algorithm gets a citizen wrong, who answers?
What G$3.6 billion actually bought
In May 2026, the acting Commissioner of Police, Clifton Hicken, gave the country a short and confident message for anyone weighing a crime: the cameras will find you. Four months later, the Ministry of Finance put a price on that promise. Its 2026 Mid-Year Report recorded G$3.6 billion spent in six months on the Safe Country Initiative, completing 193 Intelligent Video Surveillance sites across Regions 2, 3, 4, 5, 6 and 10, with networking under way on another 260. It was the largest single technology line in a public-safety budget that reached G$39.5 billion in the first half of the year.
The cameras are the visible edge of something larger. In the same May address, President Irfaan Ali described a national security system built on more than 20,000 CCTV cameras by 2030, AI crime-prediction tools, integrated national databases, and smart police stations running around the clock. He included technology that can reconstruct a vehicle's movements across multiple locations over a 72-hour window.
The announcements share a shape. Every figure describes a new capability: more cameras, more data, faster matching, wider reach. None describes who checks the system, or what happens when it points at the wrong person. Guyana is not adding a tool to an old system. It is building a new kind of state, one where a growing share of the decisions about a citizen begins with a model rather than a person.
What the four systems have in common
That last phrase, a decision that begins with a model, is the thread running through all four. An algorithmic state is a government that uses AI to make or shape real decisions about the people it governs: who is watched, who is flagged, who is diagnosed, who receives a service. When a private app gets something wrong, you close it. When a public system gets you wrong, the result can be a missed benefit, a wrongful stop, or a record that follows you.
Guyana is now running AI inside four kinds of consequential system at once, and although they look unrelated, each one does the same underlying job.
Watching and predicting
AI-enabled video surveillance across six regions, crime-prediction tools, and planned technology to trace a vehicle's route over 72 hours, all feeding a national command system.
Knowing who you are
Biometric enrolment, digital ID cards and digital passports, linked to integrated national databases designed to be interoperable with regional and international partners.
Reading your scans
AI-assisted radiology and pathology through a Mount Sinai partnership, running in 21 health facilities by mid-2026 and cutting some scan-read times from days to minutes.
Choosing where to drill
ExxonMobil told investors in July 2026 that AI helped identify four new exploration targets in the Stabroek Block, shaping the oil revenue the whole state now runs on.
The four are not equally advanced, and they are not equally risky. The health work, at least, has been paired with a public conversation about limits: in April 2026 Mount Sinai and the Ministry of Health held a seminar in Georgetown on the responsible integration of AI into care. The surveillance and identity systems have moved with far less debate. The clearest way to see what all four share is to follow a single decision, from the moment data is collected to the moment it affects a person.
Data is collected
Cameras, biometric enrolment, health records and application forms feed the system, often without the citizen being told how the data will be used.
A model scores or classifies it
Software assigns a risk score, a match, a diagnosis or an eligibility result, based on patterns in past data.
The output shapes a real decision
Who is stopped, who is treated first, who is approved, who is refused. The model does not decide alone, but it steers the outcome.
A citizen is affected
The person experiences the result. In most cases they are never told that a model was involved at all.
An error has no clear owner
If the output is wrong, no current rule in Guyana names who audits the model or who is accountable for the mistake.
Sources: letters to Kaieteur News and Stabroek News (January 2026); Guyana Data Protection Act 2023.
The first four steps are engineering questions, and Guyana is answering them quickly. The fifth is a question of law, and it is the one the country has not answered. That gap, rather than the technology itself, is the real risk Guyana now carries.
Why Guyana is only the first
It would be easy to read this as one government moving recklessly. It is closer to the opposite. Almost every Caribbean state wants what Guyana is buying; most simply cannot afford it yet. Guyana is not behaving differently from its neighbours; it is behaving the way any of them would if the same money arrived. Three conditions explain how it got here first, and why what happens next matters well beyond its borders.
Oil removed the brake
Across the region, technology is rationed by budget. A ministry wants a system, the money is not there, and the plan waits. Guyana's oil revenue removes that limit. With production targeted at roughly 1.3 million barrels a day by 2027, the state can buy and switch on advanced systems faster than it can write the rules that should govern them. The same revenue that lets Guyana lead on deployment is what lets it outrun its own rulebook.
A labour shortage that rewards speed
Guyana has a small population and a fast-growing economy, and officials openly describe a manpower gap. In a letter to Stabroek News in late 2025, one writer argued that automating administrative work is the way to run a fledgling nation short of workers, freeing its best people for higher-value tasks. That logic is reasonable, and it is also an accelerant: when automation is framed as the fix for a labour crisis, the pressure is to deploy quickly and ask governance questions later.
No rulebook was waiting
By mid-2026 Guyana had enacted the Data Protection Act 2023, but had no dedicated AI law and no adopted national AI strategy before Parliament, and reporting indicates the data-protection enforcement office was not yet fully operational. President Ali has said the time has come to build AI policy, and Minister Kwame McCoy called at the 2026 Inter-Parliamentary Union Assembly for stronger legislative frameworks and oversight, flagging algorithmic bias and AI-generated harms. Both statements describe intent rather than an enacted rule.
The template the region will inherit
Call it the template effect. Guyana currently holds the CARICOM chair and is the region's fastest-moving deployer. The vendors it selects, the systems it installs, and the standards it sets or fails to set become the reference point for smaller states that will face the same choices later with less money and less technical capacity. If Guyana builds strong oversight, that becomes the regional baseline. If it wires up surveillance and scoring systems without audit rules, that becomes the default the rest of the Caribbean imports.
Letters published in Kaieteur News and Stabroek News through 2025 and early 2026 named the gap directly, warning that facial recognition, algorithmic profiling and automated loan approvals were being introduced without a national framework defining who owns the data, who audits the algorithms, and who is accountable when errors occur. What remains unresolved in Guyana is not the technology but the accountability around it.
Three developments from 2026
None of this is hypothetical. Three developments from the past year show where the pattern already leads.
The 2026 Mid-Year Report confirmed G$3.6 billion spent on the Safe Country Initiative in six months, completing 193 surveillance sites with 260 more under way. The acting police commissioner's public warning that the cameras will find offenders captured the tone: technology as deterrent, deployed at pace, ahead of a public conversation about limits.
Mount Sinai Health System and Guyana's Ministry of Health held a seminar on integrating AI into care, described as covering foundations, ethics and decision-making. By September, AI systems were reported running in 21 health facilities. The health track shows the alternative model: adoption paired with an explicit ethics discussion, rather than adoption alone.
Stabroek News, one of the papers that carried the early warnings about unregulated AI, ceased publication in March 2026. The Associated Press framed the closure of Stabroek and Trinidad's Newsday as a blow to Caribbean democracy. As the state's capacity to collect and process data grows, the independent capacity to scrutinize it has been shrinking.
The question Guyana now has to answer
The point is not to slow Guyana down. Its lead on deployment is a real regional asset, and the April health seminar shows adoption and oversight can move together when someone insists on it. The task is narrower and more urgent: build the accountability layer while the systems are still going in, not after a wrongful flag or a leaked database forces the conversation. Six measures would close most of the gap, and none of them require slowing the technology at all.
| Action for government | What it protects against | Level |
|---|---|---|
| Publish a public register of where the state uses AI | Hidden deployment; citizens never knowing a model shaped a decision about them | Easy |
| Operationalize the Data Protection Commissioner's office | Data collected with no working body to enforce the 2023 Act | Medium |
| Require impact assessments before high-stakes AI use | Bias baked into policing, benefits and credit decisions | Medium |
| Define in law who owns citizen data and who may access it | Data captured by vendors or shared across agencies without limit | Medium |
| Mandate independent audits of surveillance and scoring systems | Wrongful flags and errors with no one answerable | Advanced |
| Set procurement rules that demand explainability and local review | Buying black-box systems no Guyanese body can inspect | Advanced |
Guyana does not have to write this from scratch. CARICOM's COTED-ICT endorsed the UNESCO Caribbean AI Policy Roadmap in July 2026, a shared framework the region has already backed. As CARICOM chair, Guyana is well placed to turn that voluntary roadmap into enforceable national rules, and to set the template the rest of the Caribbean adopts rather than importing someone else's.
How closely have you followed Guyana's AI build?
AI in Guyana: common questions
For a decade, the Caribbean's AI worry has been about falling behind. Guyana has quietly reversed the question. It is deploying faster than any of its neighbours, and the harder task now is not buying the technology but deciding, in law, who answers when an algorithm gets a citizen wrong. Caribbean AI Newsletter · September 2026
About Caribbean AI
Caribbean AI is the official directory of artificial intelligence companies, labs, and innovators in the Caribbean. We exist to connect startups, enterprises, and researchers driving the region's AI growth.
We track who is building AI in the Caribbean, what is being built, and what the region needs to compete. From Jamaica to Curaçao, from Guyana to the Bahamas, we document the people, organizations, and ideas shaping the Caribbean's AI future.