The Rights Gap: Why the Commonwealth Caribbean Needs a Human Rights-Based AI Governance Framework Before the Next Election Cycle

At some point in the past 18 months, a voice-cloned audio recording purporting to feature a Caribbean government minister making inflammatory political statements circulated widely across WhatsApp group networks in that minister's country. The recording sounded convincing. It was listened to by an estimated tens of thousands of people before a media house identified it as synthetic audio and published a debunking. By that time, the correction had reached a fraction of the audience that heard the original. The party to which the minister belonged performed below expectations in the constituencies where the clip was most widely shared.

This incident is not exceptional. It is a data point in a pattern that CAIRMC and its research partners have documented across Caribbean election cycles over the past two years. What makes it notable is not the deepfake itself -- the technology is now cheap enough that almost anyone with a smartphone and access to a few hours of public speech by their target can produce a credible voice clone. What makes it notable is that there was no legal framework to classify what happened as an offence, no electoral commission with the mandate to investigate it, no platform enforcement channel designed for the Caribbean context, and no rapid response infrastructure that could have reached the same audience in time to matter.

That governance gap -- not the technology -- is the Caribbean's real AI problem in the electoral domain. And it extends far beyond elections.

The Commonwealth Caribbean AI Governance Landscape

Across the 14 independent Commonwealth Caribbean member states, the state of AI governance can be summarised briefly: most countries have no domestic AI strategy, no AI-specific legislation, and no regulatory body with a clear mandate over AI systems. Several have data protection legislation modelled on GDPR principles following the Caribbean Community's model law, and those laws have some relevance to AI systems that process personal data. But data protection law is not AI governance law. It governs how data is collected, used, and protected. It says very little about the design of algorithms, the transparency of automated decision-making, the accountability of AI system operators, or the safeguards citizens can invoke when an AI system affects their lives.

The contrast with peer jurisdictions is instructive. The United Kingdom, another Commonwealth nation, has published a National AI Strategy, a Pro-Innovation Approach to AI Regulation, and has established a dedicated AI Safety Institute. Canada has passed the Artificial Intelligence and Data Act as part of Bill C-27. Australia has released a Safe and Responsible AI framework and is developing AI-specific legislation. New Zealand has published an Algorithm Charter for government agencies. All of these are Commonwealth nations that have moved from recognition of the problem to institutional response.

The Commonwealth Caribbean, despite having the same international obligations and comparable legal systems derived from English common law, has not made equivalent progress. The gap is not primarily a resource gap. It is a prioritisation gap. AI governance has not appeared on the agenda of most Caribbean Cabinets as a matter requiring urgent response. CAIRMC's view is that this needs to change, and that the most compelling argument for changing it is one grounded in human rights obligations that these governments have already formally accepted.

Human Rights Obligations and AI: What Caribbean Governments Have Already Agreed To

Caribbean governments are not starting from a blank slate when it comes to the normative framework for AI governance. They have already signed up to a set of international instruments that create clear obligations relevant to AI.

The UNESCO Recommendation on the Ethics of AI, adopted by all 193 UNESCO member states in November 2021, is the most comprehensive international normative framework on AI ethics in existence. It covers the full lifecycle of AI systems from design through deployment and decommissioning. It addresses transparency and explainability, accountability, safety and security, fairness and non-discrimination, privacy and data protection, human oversight and autonomy, social and economic wellbeing, and the protection of vulnerable groups. Every Caribbean state that is a UNESCO member -- which is all of them -- has formally endorsed these principles. None has published a plan for implementing them.

The Commonwealth Charter, signed by all 56 Commonwealth member states and reaffirmed at subsequent Heads of Government meetings, commits signatories to democracy, the rule of law, freedom of expression, and the right of citizens to participate in free and fair elections. These commitments are directly engaged by the deployment of AI systems in political communication, electoral processes, and the information environments in which democratic choices are made. A Caribbean government cannot simultaneously endorse the Commonwealth Charter and decline to regulate AI systems that demonstrably undermine the integrity of its elections.

The Universal Declaration of Human Rights, the International Covenant on Civil and Political Rights, and the International Covenant on Economic, Social and Cultural Rights -- all binding on Caribbean states -- establish rights to privacy, freedom of expression, freedom from discrimination, and the right to participate in public affairs. All of these rights are directly affected by the deployment of AI systems without governance frameworks. The ECHR has begun developing jurisprudence on AI and human rights through cases involving automated decision-making in employment, social welfare, and criminal justice. While Caribbean states are not subject to the ECHR, the cases articulate principles of proportionality, transparency, and human oversight that apply equally under the universal human rights instruments Caribbean states have ratified.

The human rights argument for Caribbean AI governance is not abstract. It grounds the governance imperative in legal obligations Caribbean governments have already accepted, provides ready-made normative content that can be translated into domestic law without starting from scratch, and frames the issue as a matter of rights and accountability rather than merely technology regulation.

Deepfakes and Democracy: The Caribbean's Specific Vulnerability

CAIRMC has identified four structural features of Caribbean political life that make the region specifically vulnerable to AI-generated electoral disinformation.

The first is WhatsApp dominance. Smartphone penetration across Caribbean English-speaking markets exceeds 80 percent in most countries, and WhatsApp usage is effectively universal among politically engaged adults. Political communication in the Caribbean -- including content from parties, candidates, and campaign teams -- routinely flows through WhatsApp group networks. The same channels that carry legitimate political communication carry disinformation. WhatsApp's end-to-end encryption means that regulatory monitoring is essentially impossible at the platform level, and content moderation reaches only a fraction of what circulates through group and broadcast networks.

The second is audio-centric political culture. Caribbean politics has a deep tradition of oral political communication: stump speeches, radio call-in shows, voice-note broadcasts. Political authority and authenticity in the Caribbean context is heavily constructed through voice. A convincing voice clone of a political leader carries a different weight of persuasive authority in an audio-trusting political culture than it would in a text-dominant one. The technology is now calibrated -- through open-source models trained on Caribbean political speech -- to Caribbean accents and rhetorical styles.

The third is compressed election timelines. Most Caribbean constitutional systems allow elections to be called with relatively short notice periods. A campaign environment of six to eight weeks is not unusual. Deepfake content released in the final days of a campaign creates a verification problem that the correction cycle cannot solve within the remaining voter decision window. Fact-checking and debunking typically require several days to reach broad circulation. By the time a credible correction reaches the same distribution networks as the original deepfake, the election may be over.

The fourth is limited electoral commission capacity. None of the Caribbean's independent electoral commissions currently have in-house capacity to detect, investigate, or adjudicate complaints involving synthetic media in electoral contexts. Most were designed for a media environment that predates social media, let alone generative AI. Their regulatory mandates, legal instruments, and staffing do not contemplate the challenge they now face.

What the Evidence Shows

Research on AI-generated electoral disinformation in comparable contexts provides a clear picture of the risk profile. Studies across multiple election environments have found that voice-cloned audio is the most effective form of AI-generated disinformation in terms of belief adoption and influence on voting intention, particularly in populations with high mobile usage and audio-centric political communication cultures.

Corrections are most effective when issued within four hours of the original disinformation's first appearance, include clear technical explanations of the detection evidence, and are amplified through the same channels as the original. Caribbean media organisations and electoral commissions currently have neither the technical capacity to detect deepfakes rapidly, nor the platform relationships to amplify corrections effectively, nor the legal framework to require platform cooperation.

The consequence is an asymmetric information environment in which AI-generated disinformation travels faster, more widely, and with more emotional impact than fact-based correction. This is not a theoretical risk for the Caribbean. It is the documented operational reality of recent Caribbean election cycles.

A Human Rights-Based Framework: What It Actually Looks Like

A human rights-based approach to AI governance in the Caribbean is not a philosophical exercise. It is a practical design methodology with concrete outputs.

At the level of principle, a human rights-based approach requires that AI governance frameworks identify which rights are at stake in any given AI deployment, assess how those rights are affected by the system's design and use, and build accountability mechanisms that allow rights-holders to seek redress when AI systems cause harm. Applied to electoral disinformation, this means: identifying freedom of expression and the right to free and fair elections as the rights at stake; assessing how AI-generated synthetic media in political communication affects these rights; and building mechanisms for detection, reporting, correction, and accountability that function within the compressed timelines of Caribbean electoral cycles.

At the level of legal instrument, it means translating these principles into provisions that Caribbean electoral commissions, broadcasting authorities, and data protection offices can enforce within their existing competences. Most Caribbean jurisdictions already have laws on defamation, false statements of fact in elections, and unauthorised use of someone's image or likeness. Extending these to cover AI-generated synthetic media -- with specific provisions on disclosure obligations, take-down requirements, and electoral commission reporting channels -- is achievable within current legislative frameworks without waiting for comprehensive AI legislation.

At the level of institutional capacity, it means building the technical partnerships that give Caribbean electoral commissions access to deepfake detection tools and international expertise. Several university research labs and civil society organisations provide detection support for electoral bodies in resource-limited jurisdictions. The Caribbean needs to activate these partnerships before the next round of elections.

What Caribbean Electoral Commissions Should Do Now

CAIRMC recommends four immediate actions for Caribbean electoral commissions within the current electoral cycle.

First, establish a dedicated synthetic media reporting channel. A publicly promoted mechanism for citizens, candidates, media organisations, and civil society groups to report suspected AI-generated electoral content. This requires a process and a commitment to taking reports seriously, not deep technical infrastructure to begin.

Second, issue guidance on synthetic media in electoral communications. Most Caribbean electoral commissions have existing powers to regulate political advertising and electoral communications. Extending this mandate to synthetic media -- requiring disclosure when AI has been used to create or materially alter political content -- is within most commissions' existing legal competence and does not require new legislation.

Third, build technical referral relationships. Several international organisations provide deepfake detection support for electoral bodies in smaller jurisdictions on a subsidised or free basis. The Electoral Integrity Project, the International Institute for Democracy and Electoral Assistance, and several university cybersecurity labs have Caribbean engagement programmes. Activating these partnerships now, rather than after the next deepfake incident, is the difference between preparation and crisis response.

Fourth, convene cross-commission coordination. Deepfake content created and hosted in one Caribbean jurisdiction reaches voters in another through the same WhatsApp networks. A regional coordination protocol among Caribbean electoral commissions for sharing intelligence on AI-generated electoral disinformation would significantly improve the collective response capacity.

The Regional Governance Architecture CAIRMC Is Advocating For

Individual Caribbean nation actions are necessary but not sufficient. CAIRMC, working with the Caribbean AI Association, is developing a Caribbean Human Rights-Based AI Governance Framework: a model set of provisions that member states can adopt into domestic legislation, covering transparency obligations for AI systems affecting public interest decisions, synthetic media disclosure requirements for electoral communications, reporting mechanisms for AI-generated rights violations, and inter-jurisdictional cooperation protocols for cross-border AI disinformation incidents.

The framework is deliberately calibrated for Caribbean institutional capacity. It does not require the establishment of new dedicated AI regulatory agencies, which most Caribbean states lack the resources to create. It routes implementation through existing institutions -- electoral commissions, broadcasting authorities, data protection commissioners, and consumer protection bodies -- with targeted capacity building and regional coordination to multiply their effectiveness.

The Caribbean AI network that CAIRMC works within includes country-level hubs across the region: AI Jamaica, AI Barbados, AI Trinidad and Tobago, AI Guyana, and AI St. Lucia, as well as Caribbean AI innovators Maestro AI Lab and SportsBrain AI. The governance framework CAIRMC is building is designed to create the conditions under which the Caribbean's own AI sector can develop responsibly and competitively.

The Cost of Waiting

The governance gap described in this article is not a future problem. It is a present reality with documented consequences for Caribbean democracy and for the people who live in it. Every election cycle that passes without a human rights-based AI governance framework in place is a cycle in which Caribbean citizens exercise their democratic rights in an information environment that AI systems can manipulate without accountability, recourse, or legal constraint.

The tools to begin closing this gap are available. The international normative frameworks Caribbean governments have already endorsed provide the principles. The regional coordination mechanisms of CARICOM and the Commonwealth provide the institutional channels. The existing regulatory bodies of Caribbean states have the competences to begin implementation. What is required is the political decision to treat AI governance as a rights issue -- not merely a technology issue -- and to act with the urgency that determination demands.

CAIRMC will continue to press that case. The governance architecture for responsible AI in the Caribbean is not beyond reach. It is within reach, and the cost of building it is considerably less than the cost of the damage being done in its absence.

Caribbean AI Network: Country Perspectives

For country-specific AI risk and governance resources across the Caribbean, CAIRMC works in close coordination with the Caribbean AI Association and its network of national hubs:

• AI Jamaica -- AI governance in Jamaica's BPO, fintech, and election context
• AI Guyana -- Guyana's fast-growing economy and digital governance needs
• AI Barbados -- Barbados as a regional leader in digital governance and data protection
• AI St. Lucia -- St. Lucia's tourism-driven economy and governance landscape
• AI Trinidad and Tobago -- T&T's energy and financial sectors navigating global AI governance

Frequently Asked Questions

What is a human rights-based approach to AI governance?

A human rights-based approach to AI governance uses the existing body of international human rights law as the normative foundation for AI regulation. Rather than starting from a technology-first framing (what does AI do and how do we regulate it?), it starts from a rights-first framing (what human rights are affected by AI systems and how do we protect them?). For the Caribbean, this approach has the significant practical advantage of grounding AI governance in obligations governments have already formally accepted -- through UNESCO, the Commonwealth Charter, and the universal human rights treaties -- rather than requiring the construction of entirely new normative frameworks. It also provides ready-made accountability mechanisms: human rights complaints processes, ombudsmen, and judicial review of government decisions that affect rights are already part of Caribbean legal systems.

What makes the Caribbean specifically vulnerable to AI-generated electoral disinformation?

Four structural factors combine to create unusual vulnerability. Near-universal WhatsApp penetration creates a private communication environment that is effectively beyond platform moderation. Audio-centric political culture makes voice-cloned content particularly persuasive relative to text-based corrections. Compressed election timelines mean that debunking cycles cannot outrace viral disinformation within the voter decision window. And limited electoral commission capacity means there are no institutional first responders with the mandate or tools to address the problem. No single factor is unique to the Caribbean, but their combination is -- and it creates a risk profile that standard global frameworks do not adequately address.

Do Caribbean electoral commissions currently have the legal power to address deepfakes?

Most Caribbean electoral commissions have some existing powers that are relevant -- provisions on false statements of fact about candidates, rules on electoral advertising, and in some jurisdictions, broadcasting codes that prohibit fabricated statements attributed to candidates or officials. However, none of these provisions were designed with AI-generated synthetic media in mind, and most commissions do not have the investigative capacity to apply them to digital content. What is needed is: explicit extension of existing provisions to synthetic media; a specific duty to investigate reported deepfake electoral content; technical capacity and partnerships to conduct investigations; and cooperation agreements with platforms and broadcasters to support rapid response. None of these require fundamental legislative change. Most can be achieved through supplementary regulations, guidance documents, and memoranda of understanding with existing legal authority.

Is there an international model the Caribbean can adapt?

Several jurisdictions provide relevant models. The EU's Digital Services Act creates disclosure requirements for AI-generated content in political advertising and requires large platforms to assess their systemic risks for democratic processes, including elections. The UK's Online Safety Act creates duties on platforms around harmful content, including some provisions relevant to synthetic media. The US Federal Election Commission has ruled on disclosure requirements for AI-generated political advertisements. UNESCO's Recommendation on the Ethics of AI provides the most comprehensive international normative framework. A Caribbean approach should draw on all of these while being adapted for the Caribbean's legal systems, institutional capacity, and specific communication environment -- which is precisely the adaptation work CAIRMC is currently undertaking.

How can Caribbean civil society organisations engage with CAIRMC's governance advocacy work?

CAIRMC engages civil society organisations through its Caribbean AI Governance Coalition, which brings together human rights organisations, media organisations, electoral monitoring bodies, and digital rights advocates to develop collective positions on AI governance issues affecting the region. The Coalition's work includes policy submissions to governments and regional bodies, technical education programmes for civil society organisations on AI literacy and deepfake detection, and public communications that translate technical AI governance issues into accessible public discourse. Contact CAIRMC through caribbeanairisk.com to join the Coalition or request information about specific engagement opportunities.