Legal

AI Charter

Governing the use of artificial intelligence across The Sovereign Press, Caribbean Compass, and The Sovereign Standard

Effective: 29 June 2026Last Updated: 29 June 2026

About this Charter

▸Version 1.4 — Approved by the Principal, 29 June 2026 (corpus-aligned; benchmarked against international consulting responsible-AI practice, academic-publishing AI policy, and Caribbean regional AI instruments; EU AI Act adopted as external development standard).
▸Scope: All publications and analytical deliverables issued under FCG and its imprint, The Sovereign Press.
▸Issuing authority: Robert D. Forrester, Principal, Forrester Consulting Group; Chairman & CEO, The Forrester Group (Guyana), Inc.
▸Review cycle: Annual, with out-of-cycle revision on regulatory or architectural change.

Preamble

Forrester Consulting Group produces books, indices, sector reports, and advisory deliverables for Caribbean governments, multilateral institutions, the regional private sector, and the reading public. We use artificial intelligence in the production of that work. We have used it openly and as a matter of method, not exception. We will continue to use it.

We say so without apology and without spectacle. A small firm working from Georgetown does not have the option of pretending the contest of tools passes it by; the more deliberate posture is to make the tool pass through a discipline of our own design, so that the using of it confers rigour rather than eroding it. That is the same instinct that runs through the firm's analytical work — the conviction that position is made, not inherited, and that a small operation earns its standing by the exactness of its method rather than the size of its hand. This charter states the method as it bears on AI.

The discipline described here is not aspirational. It is already built into how the firm works: into the sequenced production routines that govern how a manuscript moves from draft to print and how a sector report moves from research to ministerial delivery; into the verification layers that close every fact against a primary source; into the consistency audits that run before a multi-document programme is released; into the voice standard that keeps the firm's writing recognisably its own. The charter does not invent obligations. It names, for the public, the standards the firm's own architecture already enforces.

We make these commitments publicly because they are real obligations owed to real readers and clients. They apply to the Principal, to FCG, to The Sovereign Press, and to anyone who publishes under either imprint. We revisit them each year, and sooner if the law, the technology, or our own way of working shifts enough to open a gap.

We publish this charter for a plain reason. FCG's work is cited in Cabinet, submitted to university presses, and read by a diaspora that trusts the voice on the page. That trust rests on a standard. We would rather set the standard down in the open than let anyone guess at it.

Article I — Human Responsibility and Authorial Authority

1.1 Every work published by FCG or The Sovereign Press is the responsibility of a named human author or, for institutional deliverables, of the Principal. Nothing in our use of AI dilutes that responsibility. Final accountability for AI governance across FCG and The Sovereign Press rests with the Principal, Robert D. Forrester, who holds the standard described in this charter and answers for its application.

1.2 No work is published without a human reading and approving it first. For book-length and ministerial-grade works, that human is the Principal or the named author. The firm's production routines are built so that human approval is a required gate, not a courtesy — a deliverable does not complete its sequence until a person has signed off on it. This is the human-in-the-loop standard recognised across both serious advisory practice and scholarly publishing: a person reviews, edits, and takes responsibility for every output, and it is the abandonment of that responsibility, never the use of a tool, that would constitute a failure of integrity.

1.3 Authorship at FCG means origination of the intellectual work. The author originates the thesis, the central argument, the analytical architecture, the framework, and the conclusions of every work published under their name. AI does not originate these, and AI is not, and cannot be, an author of any FCG or Sovereign Press work; authorship requires accountability, and only a person can carry it. The test is not how much AI touched the work but what it was permitted to decide. The author settles what the work argues, claims, and concludes before any tool is brought in to help express it. A tool may sharpen the sentence; it may not choose the position.

1.4 Authorial voice. The authorial voice in FCG's literary and scholarly work is the Principal's own — formed over thirty years of practice, argument, and writing, and rooted in a distinct Caribbean intellectual tradition. AI does not generate that voice and does not stand in for it. Where AI assists with drafting, the firm holds the assisted text to an explicit voice standard built from the Principal's own body of work, so that assistance conforms to the author's established voice rather than displacing or diluting it. The firm treats the integrity of that voice as an asset to be protected and deliberately strengthened over time: the standard is maintained, refined, and applied precisely so that the work on the page remains the author's in cadence and judgement as well as in substance. The voice is not a product of the tool. The tool is held to the voice.

1.5 AI is never the sole verifier of a published fact. A human makes the final call on whether a fact is true and may be published. This holds whatever part AI played in researching, drafting, or assembling the work.

1.6 Byline and attribution are honest. Where Robert D. Forrester is named as author, the intellectual work is his. Where a work is issued institutionally under the FCG name, the Principal answers for it. FCG does not publish work under a human byline that the named human did not intellectually originate and approve.

Article II — The Limits on AI Use at FCG

2.0 Affirmative standard. Beyond the limits that follow, FCG commits affirmatively to accuracy and reliability in its AI-assisted work. The firm uses AI to raise, not lower, the standard of what it publishes: to widen research, sharpen analysis, and tighten verification. Where AI assistance cannot meet the firm's standard of accuracy for a given task, the task is done without it. Trustworthy output is the object; the tool is judged against it.

2.1 Permitted functions. Within FCG's production cycle, AI assists with: research and source identification; structuring and drafting of author-originated material; fact-checking and verification support; data analysis, indexing, and visualisation; citation formatting; language refinement and copy editing; typesetting and production composition; translation support; distribution and audience copy. Its role is to help carry out work the author has already conceived.

2.2 Origination boundary. AI does not originate the argument, thesis, analytical framework, or conclusions of any FCG work. For literary works, it does not originate the authorial voice, the narrative, or the lived testimony. For scholarly works, it does not originate the central scholarly claim or its defence. The firm's frameworks, doctrines, and indices are human intellectual property, developed and held by the firm; AI helps articulate and apply them, and never authors them.

2.3 No fabricated facts, sources, or citations. AI is never the origin of a fact, statistic, quotation, or citation that enters a published FCG work unverified by a human against a primary or authoritative source. Material that cannot be verified is flagged for primary-document resolution and held back until it is verified or removed. This is not a new promise made for this charter; it is the standing discipline already applied to the firm's manuscripts and indices, where unconfirmed items are quarantined as a matter of routine rather than waved through.

2.4 No fabricated attribution. AI never generates a quotation attributed to a real person, nor commentary written to imitate the voice of an identifiable real person, that is then presented as that person's words. Quotations from real persons in FCG work come from documented, human-verified sources.

2.5 No synthetic depiction of real persons or events. AI never generates a photorealistic image of a real person or real event presented as a record of fact. Illustrative or conceptual graphics, cover art, and diagrams may be AI-assisted only where they do not depict a real person or event in a way that could be mistaken for documentary record, and where they are marked as illustration. Any AI-generated illustration carries machine-readable marking identifying it as such, consistent with EU AI Act Article 50 (effective August 2026) where applicable.

2.6 No fabrication or misrepresentation. FCG does not publish fabricated data, invented findings, deepfakes, or any content engineered to deceive. On these points there is no discretion and no situational exception.

2.7 No false certainty in advisory work. Where FCG advisory deliverables make claims that bear on government or institutional decisions, AI-assisted analysis is held to the firm's evidence-grading discipline. Claims are weighted by confidence and source quality, and AI-generated inference is never dressed up as verified fact. A reader in a ministry is entitled to know the difference between what the firm has established and what it has reasoned toward, and the firm's analytical method is built to keep that line visible.

2.8 Fairness and representativeness. FCG's analytical work concerns the Caribbean and its position in a wider order, and the firm is alert to the ways AI tools — trained predominantly on metropolitan sources — can underweight, flatten, or misread Caribbean realities. The firm treats this as a substantive analytical risk, not a procedural footnote. Sources, framings, and findings are checked for representativeness across the territories, languages, and sectors the work addresses, and AI-assisted output is corrected where it imports a metropolitan default in place of the Caribbean particular.

Article III — The Tiered Production Standard

FCG produces three distinct categories of work, each moving through its own architected production sequence. Each carries a calibrated standard. The origination boundary in Article II applies to all three without exception.

3.1 Literary and scholarly works. (e.g. Don't Just Leave, Coming Home, The Rice Coast, A Caribbean Plutarch, Governing the Black Box, Definitions, the Go Strategies series.) The named author originates the argument, voice, and intellectual architecture. These works move through the firm's chapter-cycle production routine, in which drafting, editorial deepening, citation verification, and voice checking are sequenced steps, not afterthoughts. AI assists with research, structuring, citation, editorial refinement, and production. For academic-track works, every source is verified against the primary document before publication, and any source that cannot be confirmed is removed. Academic-track disclosure: where an FCG or Sovereign Press work is submitted to a university press, a peer-reviewed journal, or any venue with its own AI-disclosure requirement, the firm complies with that venue's requirement in full, including any obligation to declare and explain AI assistance in the work itself. This charter's firm-level disclosure position (Article VI) operates alongside, and never in substitution for, a publisher's specific declaration rules. The author's moral right to be identified as author of the work is asserted and is genuine.

3.2 Advisory and analytical deliverables. (Sector reports, indices, ministerial briefs, diagnostics, financial models.) The Principal originates the analytical frame and answers for the findings. These works move through the firm's report and ministerial production routines, which carry their own verification gates: facts and figures are checked against authoritative sources; indices and models are recalculated and error-checked rather than assumed correct; the firm's verified data registry for recurring Caribbean and Guyana figures is consulted before release; and known corrections are propagated across the affected suite. AI assists with research, data analysis, modelling, drafting, and visualisation, inside those gates.

3.3 Recurring publications. (Caribbean Compass, The Sovereign Standard.) The Principal originates the analysis and editorial position. These move through the firm's recurring-publication routines, which apply the same fact-verification and voice standards as the firm's longer work, scaled to cadence. AI assists with research, drafting support, and production. These publications carry FCG's analytical voice and the Principal's editorial responsibility.

Article IV — Sources, Verification, Corrections, and Accountability

4.1 Every material fact in every FCG publication is sourced and verified by a human. Whether or not a tool was used along the way, the Principal answers for the sourcing under the responsibility set out in Article 1.1.

4.2 FCG does not recompose another author's or journalist's work without primary-source verification and full attribution to the original author and outlet. The firm's citation discipline treats attribution as a first-order obligation, applied uniformly whether a passage was researched by hand or with assistance.

4.3 FCG maintains a verified registry of recurring Guyana and Caribbean data — its factual anchors. Published claims are checked against this registry before release, and where the registry and a new source disagree, a human resolves the conflict before anything goes out. The registry is the firm's institutional memory of what it has established to be true, and it is consulted by design.

4.4 When an error appears in an FCG publication, it is corrected under one protocol regardless of cause. Material corrections are logged in an FCG corrections record. Where an error traces to unverified AI output, the production step that allowed it is reviewed and tightened, so that the failure is closed at the level of the routine and not merely at the level of the single document. The firm treats its own prior corrections — including documented corrections to recurring factual anchors — as evidence that the verification discipline is real and working, not as embarrassments to be hidden. A firm that publishes its corrections is a firm that can be trusted to have a verification standard at all.

4.5 FCG keeps an internal record of AI use noting, for substantive deliverables, the nature of the assistance and the human reviewer who approved the work. This record is internal to firm operations and supports the accountability commitments in this charter.

4.6 Consistency is audited, not assumed. Before a multi-document programme is released, the firm runs a cross-deliverable consistency check across naming, framework reference, factual anchoring, citation, terminology, and voice, so that a suite of related documents speaks with one settled position rather than several drifting ones. This audit is part of why the firm can stand behind a programme as a whole and not only its pieces.

Article V — Data Discipline, Confidentiality, and Vendors

5.1 FCG uses AI vendor services only where the vendor's data handling is consistent with the Principal's confidentiality obligations to clients and where the vendor is authorised for the work in question. The firm exercises ongoing diligence over the tools it relies on — what they do with input, where data rests, and how their terms bear on confidentiality and intellectual property — and reconsiders that reliance as tools and terms change.

5.2 The following never pass through any third-party AI system uncontrolled, regardless of vendor: client confidential information; unpublished documents shared in confidence; primary documents under legal privilege or seal; commercially sensitive financial data held in confidence; personal data protected under Guyana's Data Protection Act 2023 or applicable CARICOM data protection law; material under embargo.

5.3 FCG's advisory engagements may carry client-specific data-handling terms stricter than this charter, and clients and funders — including multilateral institutions with their own AI procurement and governance frameworks — may impose requirements of their own. Where any such requirement is stricter than this charter, it governs for that engagement. The firm's confidentiality obligation to a client is never weakened by a general policy; it is only ever strengthened by a specific one.

5.4 FCG complies with Guyana's Data Protection Act 2023 and monitors CARICOM data-protection developments and EU AI Act enforcement affecting FCG-deployed AI systems. The controller for FCG-held personal data is The Forrester Group (Guyana), Inc.

Article VI — Disclosure and Governance

6.1 Position on disclosure. FCG does not attach a separate AI-use label to each work it publishes. Our reasoning is twofold. First, a label on every page treats the tool as a special event long after it has become ordinary equipment in law, finance, scholarship, and publishing — the same way a serious firm does not stamp each document to announce that it was produced on a computer. Second, the disclosures that actually matter to a reader — who is accountable, where the facts came from, whether the author truly wrote the work — are commitments this charter already makes at the level of the firm and enforces through the firm's production architecture. A per-page label would add noise without adding accountability. The charter, backed by that architecture, is the disclosure. This firm-level position does not override the specific disclosure rules of any venue to which the firm submits work; where a publisher or journal requires a declaration, the firm makes it (Article 3.1).

6.2 Answering the direct question. Anyone may ask whether a particular FCG work involved AI, and the firm answers plainly: yes, AI assisted in researching and producing it; the thinking, the writing or direction of it, the verification, and the final approval were the named author's and the Principal's. The charter sets out where that assistance was permitted and where it was not.

6.3 Canonical publication. This charter is the canonical statement of FCG's AI practice. It is published on the FCG platform and indexed so that automated systems referencing FCG work can find it directly. It is linked from the FCG About page.

6.4 Keeping the charter current. The Principal reviews this charter at least once a year. Small clarifications are made on the Principal's own authority. Three kinds of change are weightier — touching the Article II prohibitions, the charter's scope, or how the charter itself is governed — and the Principal takes legal advice before making them. The annual cycle is a floor, not a ceiling: if a change in the law, the adoption of a new system, or a mistake the firm has had to correct exposes a problem the document does not yet address, the Principal amends it without waiting.

6.5 Who is bound. Anyone who publishes under the FCG or Sovereign Press name works within this charter — in particular the origination limit in Article II and the verification requirement in Article IV. Contributors are expected to know its terms before their work appears, and the firm's drafting routines apply its standards as a matter of course.

6.6 External standard. FCG adopts the European Union Artificial Intelligence Act (Regulation (EU) 2024/1689) as its external development standard. Although FCG is not established in the European Union, the firm holds its own AI development and use to the Act's risk-based discipline as a matter of policy: AI uses prohibited under the Act are not undertaken by FCG; where the firm's work touches uses the Act treats as high-risk, the firm applies the Act's corresponding expectations of human oversight, transparency, record-keeping, accuracy, and data governance; and the firm meets the Act's transparency obligations for AI-generated and AI-assisted content, including the machine-readable marking of synthetic media required under Article 50 from August 2026. The firm treats the EU AI Act as a floor it builds upon, not a ceiling, and reviews its development practice against the Act as the Act's provisions phase into force. This commitment is a standing one: FCG develops and uses AI in conformity with the EU AI Act going forward, and the Principal logs and maintains it as firm policy.

Article VII — Intellectual Property and the Training of AI Systems

7.1 FCG’s work is human intellectual property. The firm's books, indices, frameworks, doctrines, manuscripts, and the Principal's body of written work are original human intellectual property, owned by The Forrester Group (Guyana), Inc., with the Principal's moral rights asserted. AI assistance in producing a work does not change its ownership or its human authorship.

7.2 Ownership of AI-assisted output. Where AI assists in producing an FCG work, the resulting work — its argument, structure, expression, and final form as approved by the author — is owned by The Forrester Group (Guyana), Inc. under the same terms as any other FCG work. The firm does not publish, and does not claim ownership of, raw unedited machine output; what the firm publishes is human-authored work that AI helped produce.

7.3 Position on the use of FCG’s work to train AI systems. FCG's published and unpublished work is not made available for the training of third-party AI systems except by the firm's express, written, opt-in consent on terms the firm controls. This applies to the Principal's literary and scholarly corpus, the firm's frameworks and indices, and client-confidential material without exception. Where a publishing, distribution, or platform partner proposes to license FCG content for AI training, the firm treats that as a distinct decision to be made deliberately and in the interest of the author and the firm — never as a default buried in a standard agreement. Client-confidential material is never available for AI training under any terms.

7.4 The author’s voice and corpus. The firm regards the Principal's authorial voice and corpus as among its most valuable assets and protects them accordingly — both from dilution in production (Article 1.4) and from unconsented use in the training of external systems (Article 7.3). The firm's interest is in strengthening and preserving that distinct point of view, not in surrendering it to any system the firm does not control.

Article VIII — Standards Alignment

This charter is written in FCG's own terms, but its commitments are deliberately consistent with the standards now expected of serious institutions in the firm's fields of work — in international advisory practice, in scholarly and literary publishing, and in the Caribbean region itself.

In advisory and analytical practice, the charter reflects the principles common to responsible-AI frameworks used by leading professional-services firms: human oversight across the work; an affirmative standard of accuracy and reliability; attention to privacy, confidentiality, and intellectual property in the data used; diligence over third-party tools; and clear accountability vested in a named person.

In scholarly and literary publishing, the charter reflects the consensus now settled across major academic publishers: that AI cannot be an author; that the human author is accountable for the accuracy, integrity, and originality of the work; and that AI assistance is declared where the venue requires it. FCG meets these expectations as a matter of course and complies with the specific AI policies of any press or journal to which it submits work.

And in its own region, the charter aligns with the Caribbean AI framework now taking shape — set out in full in Article IX — which the firm treats as authoritative for its conduct and helps to advance. The firm states these alignments not to borrow authority from others but to make plain that its own discipline, arrived at in its own way, meets the bar that the most serious institutions in all three settings now set — and in several respects sets it higher.

Article IX — Alignment with Caribbean Regional Standards

FCG is a Caribbean firm, and its AI practice is built to meet the standards the region is setting for itself — not only those set in metropolitan centres. The firm reads the emerging Caribbean AI framework as authoritative for its own conduct and aligns to it deliberately.

9.1 Regional principles. FCG aligns its AI practice with the guiding principles articulated for the Caribbean by the Caribbean Telecommunications Union's Caribbean Artificial Intelligence Task Force — human-centric and rights-based AI; inclusivity, equity, and sustainable development; data sovereignty, governance, and security; regional collaboration and harmonisation; and resilient innovation. The firm regards these as the regional expression of the same commitments this charter already makes, and supports the harmonisation of AI governance across CARICOM rather than a fragmented, jurisdiction-by-jurisdiction approach.

9.2 Regional ethics instruments. FCG aligns with the UNESCO Recommendation on the Ethics of Artificial Intelligence and with the UNESCO Caribbean Artificial Intelligence Policy Roadmap, including its emphasis on cultural identity, the protection of intellectual property, human rights, and education. The firm regards the protection of Caribbean cultural and intellectual production in the age of AI as a matter of regional interest, not merely a private one.

9.3 Editorial and authorship integrity, regional precedent. As a publisher and advisory firm, FCG looks to the strongest Caribbean precedents on AI, authorship, and human responsibility — in particular the Caribbean Court of Justice's Practice Direction No. 1 of 2025 on the use of generative AI, and the Caribbean Examinations Council's Responsible Generative AI Policy Framework. The principles those instruments establish — full human responsibility for AI-assisted output, independent verification of AI-generated material, disclosure of AI use where required, the prohibition on entering confidential or privileged material into open AI tools, and the rule that the substance of an authored work may not be machine-generated — are principles FCG already applies and restates here as its own.

9.4 Data sovereignty and protection. FCG treats Caribbean data sovereignty as a substantive commitment. The firm complies with Guyana's Data Protection Act 2023 and holds itself, as a matter of policy and regardless of any single jurisdiction's enforcement capacity, to the stronger data-protection practice now emerging across the region — including the right of a person not to be subject to a solely automated decision without human intervention, purpose limitation, data minimisation, and safeguards on cross-border transfer. Client and personal data governed by Caribbean data-protection law is handled accordingly, and is never surrendered to the training of external AI systems (Article 7.3).

9.5 A standards-maker, not only a standards-taker. FCG holds that the Caribbean should help author the standards that govern AI in the region rather than inherit them wholesale from elsewhere. The firm offers its own practice — this charter, its verification and authorship discipline, and its analytical work on Caribbean digital and AI policy — as a contribution to that regional effort, and engages the institutions building the regional framework in that spirit. Caribbean organisations gauging their own readiness to adopt AI can begin with the firm’s free Caribbean AI Maturity Diagnostic, a structured self-assessment of AI maturity.

Signed

Robert D. Forrester — Principal, Forrester Consulting Group; Chairman & CEO, The Forrester Group (Guyana), Inc.

Forrester Consulting Group is a member of The Forrester Group (Guyana), Inc.

Publishing imprint: The Sovereign Press.

Email: robert@forresterconsultgy.com

Telephone: +592-683-5373