New framework asks whether nations can keep information trustworthy in the age of synthetic media

Generative artificial intelligence has made it cheap and fast to fabricate convincing text, images, audio, and video, and the consequences now reach far beyond ordinary misinformation. A new Perspective article published in Frontiers in Artificial Intelligence argues that the world’s prevailing AI readiness frameworks measure the wrong thing for this emerging era: they assess whether states can adopt, deploy, and regulate artificial intelligence, but they do not systematically evaluate whether societies can preserve the authenticity and trustworthiness of the information circulating within AI-enabled environments. The authors, a team of Somali researchers affiliated with Somali International University, Benaadir Research, Consultancy and Evaluation Centre, and Somali National University, introduce a complementary governance construct they call Information-Integrity Readiness, or IIR, and illustrate it through Somalia’s rapidly digitizing but institutionally fragile landscape.

The central problem the authors identify is what they term the Information-Integrity Gap. Synthetic media undermines a long-standing assumption that audiovisual content provides reliable evidence of reality, threatening the credibility of journalism, science, governance, and democratic decision-making. The article describes what it calls the Generative AI Paradox: as fabricated content, identities, and interactions become harder to distinguish from authentic communication, widespread availability of synthetic media may itself reduce confidence in digital evidence, raising the institutional and societal cost of establishing what is true. Existing readiness indices emphasize technological infrastructure, innovation capability, digital skills, regulatory maturity, and institutional preparedness, yet these dimensions do not determine whether media can be authenticated, whether provenance can be verified, or whether citizens can continue to rely on digital evidence once synthetic content becomes pervasive.

Because the work is a conceptual Perspective rather than an empirical study, its methodology is an interdisciplinary narrative synthesis of scholarship on AI governance, readiness, digital resilience, misinformation, and synthetic media, drawing primarily on literature from 2021 to 2026 alongside earlier foundational theory. The authors are explicit about the limits of this approach: they claim no exhaustive retrieval of sources, no formal study-quality scoring, and no empirical validation of the resulting framework. Instead, the synthesis proceeded in stages, first identifying the dominant object of assessment in existing readiness frameworks, then mapping synthetic-media risks that object fails to capture, such as provenance failure, identity deception, evidentiary uncertainty, and declining public confidence, and finally clustering recurring governance capacities into an integrated architecture.

That architecture consists of five interdependent pillars. Institutional readiness covers legitimate ownership, coordination, accountability, and incident response across public and private actors. Technical authenticity readiness covers authentication, provenance standards, forensic verification, interoperability, and trusted identity mechanisms. Legal and regulatory readiness covers enforceable duties, platform obligations, remedies, evidentiary rules, and rights safeguards. Societal resilience readiness covers literacy, awareness, trusted communication, inclusion, and the public’s interpretive capability. Sector-specific readiness covers domain protocols, exercises, escalation thresholds, and continuity arrangements in areas such as health, elections, finance, education, and national security. The authors argue each pillar performs a non-substitutable function, so weakness in one reduces the effectiveness of the others, making the set a minimum complete architecture rather than a menu of optional interventions.

The framework is anchored in three theoretical traditions that serve as cross-cutting explanatory lenses. Institutional theory explains why formal rules and organizations are insufficient on their own: a data-protection authority or media regulator may exist on paper while responsibilities remain fragmented and enforcement inconsistent. Policy capacity theory distinguishes the analytical capacity to diagnose risks, the operational capacity to mobilize organizations and resources, and the political capacity to sustain legitimacy and support, warning that policy failure can result from imbalance among these even when institutions and technologies are present. Socio-technical systems theory explains why provenance metadata, for example, has limited value if platforms do not preserve it, courts do not recognize it, journalists lack verification routines, or citizens cannot interpret authenticity signals.

Applying the framework to Somalia yields an illustrative, deliberately provisional assessment rather than a national ranking. The authors note that Somalia has expanded mobile connectivity, mobile money, digital identity initiatives including e-Aqoonsi and HUBIYE, and e-government services, and has established a Data Protection Authority, SOMCIRT, a Data Protection Act in 2023, and a Cybersecurity Law in 2026, alongside a 2025 Central Bank requirement for national identification in banking transactions. Yet dependence on foreign data hosting, limited AI-specific regulation, uneven implementation capacity, and the absence of dedicated synthetic-media governance constrain the country’s ability to preserve information authenticity. Using transparent qualitative maturity criteria, the assessment assigns Emerging status to the institutional, technical, legal and regulatory, and societal pillars, and Nascent status to sector-specific readiness, where protocols and escalation arrangements are largely undocumented.

The diagnostic value, the authors contend, lies in exposing a blind spot: conventional AI readiness can register visible progress in connectivity, identity infrastructure, and digital services while overlooking whether institutions can establish provenance, coordinate authenticity disputes, communicate verified evidence, and manage synthetic-media incidents. Somalia is not presented as an exceptional crisis context but as an illustrative case in which the divergence between digital adoption and governance capacity is particularly visible. Notably, Somalia is absent from major global AI readiness assessments, a pattern the authors say reflects how fragile states frequently fall outside prevailing evaluation frameworks despite heightened vulnerability to synthetic-media harms.

To test transferability, the article offers Kenya only as a concise illustration, not a comparative case. Kenya has a more mature AI-policy ecosystem, including a national AI strategy and evolving legislation, yet continues to face deepfakes, misinformation, crisis-communication challenges, and uneven digital literacy. Research on deepfake regulation in Kenya has emphasized needs for forensic expertise and public awareness, concerns the authors map onto the technical, legal, societal, and sector-specific pillars. The claimed portability is deliberately modest: IIR can serve as a common diagnostic architecture across settings when its indicators and thresholds are adapted to national context, but the Kenya section establishes no comparative performance or causal claims.

The Perspective advances four theory-derived propositions explicitly framed for future empirical testing rather than as findings. These hold that IIR should be empirically distinguishable from AI readiness; that balanced capacity across the five pillars should be associated with stronger resilience to synthetic-media incidents than excellence in any single pillar; that the effect of technical authentication mechanisms should depend on institutional coordination, legal enforceability, and societal capacity to interpret verification signals; and that sector-specific readiness may mediate the translation of national governance capacity into operational protection in high-risk domains. For measurement, the authors recommend distinguishing formal capacity, implementation capacity, and demonstrated performance, and favor a profile or dashboard approach over a single composite score, so that severe weakness in one pillar remains visible rather than being offset by high scores elsewhere.

The article closes with a phased policy roadmap: short-term priorities include assigning institutional ownership, strengthening digital literacy, and developing Somali-language verification tools; medium-term priorities focus on provenance infrastructure, legal modernization, and cross-sector coordination; long-term priorities emphasize adaptive regulation, regional cooperation, and recurring preparedness exercises. The authors stress that every outcome remains a theoretical expectation pending validation through indicator development, expert review, pilot testing, and comparative studies. Their broader argument is that the next generation of AI governance must move beyond asking whether states are ready to adopt and regulate artificial intelligence, and must instead ask whether they possess the institutions, capacities, and socio-technical arrangements needed to preserve the authenticity of the information on which public trust, democratic integrity, and evidence-based decisions depend.

Subject of Research: A conceptual governance framework for national readiness to preserve information integrity against synthetic media, applied illustratively to Somalia

Article Title: Beyond AI readiness: an information-integrity readiness framework for governing synthetic media in Somalia

Article References: Ahmed, A. A., Mohamed, A. A., Ali, M. K., Mohamed, Y. H., Omar, M. A., Mohamed, A. A., & Hersi, I. M. (2026). Beyond AI readiness: an information-integrity readiness framework for governing synthetic media in Somalia. Frontiers in Artificial Intelligence, 9, 1959660. https://doi.org/10.3389/frai.2026.1959660

Image Credits: AI Generated

DOI: 10.3389/frai.2026.1959660

Keywords: information integrity, synthetic media, AI readiness, deepfakes, Somalia, AI governance, digital resilience, provenance, policy capacity, socio-technical systems, fragile states, misinformation