AI, Security, and Intelligence: Open and Covert Systems

Dr Maitland Hyslop1 January 2026

Abstract

Artificial Intelligence (AI) is transforming intelligence from a human-centred craft into a computational infrastructure embedded across security, defence, law enforcement, and governance systems. Intelligence is no longer primarily interpretative but algorithmic: predictive, automated, and continuous. This shift has democratised open-source intelligence while simultaneously enabling covert manipulation at unprecedented scale. The same technologies that empower civil society also facilitate deepfakes, psychographic warfare, automated espionage, and cognitive operations. As a result, security challenges now extend beyond physical and cyber domains into perception itself. This article examines AI across open and covert intelligence ecosystems, evaluates risks to critical infrastructure and civil liberties, and analyses emerging governance frameworks including ISO/IEC 42001, the NIST AI Risk Management Framework, and the EU AI Act. It argues that intelligence governance must evolve from algorithm oversight to systems governance—embedding transparency, human control, resilience, and international norms to stabilise an accelerating algorithmic arms race.

Keywords: artificial intelligence, intelligence studies, cybersecurity, OSINT, governance, algorithmic warfare, surveillance, national security

1. Introduction: The New Intelligence Paradigm

Artificial Intelligence has fundamentally redefined intelligence. Where intelligence once depended on human judgment, espionage networks, and interpretive analysis, it is now increasingly computational: a fusion of satellite imagery, behavioural analytics, signal interception, and algorithmic inference. Spycraft has become datacraft.

Modern intelligence ecosystems process heterogeneous data streams at speeds and scales impossible for human analysts. Machine learning enables predictive modelling, anomaly detection, and real-time threat assessment. Intelligence is therefore no longer episodic or retrospective; it is adaptive, continuous, and embedded within cyber-physical infrastructures.

AI now permeates the entire intelligence spectrum. Open-source intelligence (OSINT) uses natural language processing and computer vision to analyse public information flows. Signals intelligence (SIGINT) relies on machine learning to detect subtle anomalies in communications and network traffic. Covert operations deploy generative systems to conduct disinformation, impersonation, and automated surveillance. Intelligence has become a living system rather than a discrete function of the state.

This transformation creates a core dilemma: the same computational power that enhances security also magnifies manipulation and control. Governance must therefore address intelligence not simply as technology, but as infrastructure with systemic societal consequences.

2. Open Intelligence and the Democratisation of Analysis

AI has radically lowered the barriers to intelligence work. Tasks once reserved for state agencies—geolocation, satellite analysis, pattern recognition—are now accessible to journalists, researchers, and citizens with cloud computing resources. Organisations such as Bellingcat and the Centre for Information Resilience demonstrate how AI-enhanced OSINT can verify military movements, document human rights abuses, and counter disinformation.

This democratisation has two opposing effects.

First, it empowers accountability. Transparent, evidence-based investigations can challenge official narratives. Bellingcat’s analysis of the MH17 missile system showed that publicly available data could outperform classified processes in both speed and credibility.

Second, it introduces instability. Generative AI enables deepfakes, fabricated evidence, and automated propaganda at scale. Authenticity becomes contested. When synthetic media becomes indistinguishable from reality, trust collapses.

Verification technologies therefore become foundational to democratic intelligence. Blockchain-based provenance systems, watermarking standards, and AI-driven detection tools aim to authenticate origin and integrity. Yet each approach has limitations—metadata can be stripped, watermarks removed, detection models evaded. No single solution suffices.

The challenge is thus trust calibration: maintaining openness while ensuring verifiability. Governance must protect the benefits of democratised intelligence without enabling epistemic chaos.

3. Covert Intelligence and the Algorithmic Arms Race

If open intelligence empowers transparency, covert AI reshapes perception itself. Modern intelligence competition increasingly operates in the cognitive domain, where influence replaces force and narratives become weapons.

Deepfake technologies destabilise political trust. Psychographic profiling enables precision manipulation of behavioural vulnerabilities. Automated cyber-espionage scales intrusion faster than human defenders can respond. Generative propaganda floods information ecosystems with persuasive falsehoods. Strategic advantage lies not in destruction but in shaping how reality is perceived.

These capabilities produce an algorithmic arms race. States compete to predict, pre-empt, and manipulate events before adversaries react. Information asymmetry becomes the decisive asset.

The implications resemble earlier nuclear or cyber competitions but with a critical difference: the battlespace is cognition. Errors propagate through societies rather than battlefields. Escalation may occur through misinformation rather than missiles.

Countermeasures—provenance standards, explainability requirements, human-in-the-loop oversight, and collaborative intelligence sharing—seek to slow this escalation. Yet the pace of innovation often outstrips governance capacity. Intelligence leaders must therefore operate simultaneously across human, digital, and perceptual domains, balancing speed with ethical restraint.

4. AI and the New Cyber Battlespace

Cybersecurity has expanded beyond infrastructure defence to the defence of cognition. AI systems now autonomously detect intrusions, adapt to threats, and sometimes initiate countermeasures. However, adversarial AI systems simultaneously learn to deceive those defences, creating autonomous attack–defence cycles.

This raises difficult questions of accountability. When an AI system initiates or responds to a cyberattack, responsibility becomes diffuse. Legal doctrines premised on human agency struggle to assign liability.

Critical infrastructure dependence compounds these risks. Hyperscale cloud providers underpin much of the global AI ecosystem, creating systemic single points of failure. Manipulated training data or adversarial inputs can cause catastrophic misclassification in military, financial, or medical contexts.

Containment therefore becomes a governance priority. AI-enabled defence systems must remain explainable, auditable, and subject to ultimate human control. Frameworks such as ISO/IEC 42001 and the NIST AI Risk Management Framework aim to embed lifecycle risk management, continuous monitoring, and human oversight. These mechanisms seek not to eliminate autonomy but to prevent uncontrolled escalation.

5. National Security and Critical Information Infrastructure

AI has fused national security with critical information infrastructure. Energy grids, financial systems, telecommunications, and cloud platforms are now inseparable from defence capability. Data sovereignty increasingly equates to national sovereignty.

Governments respond by institutionalising AI governance frameworks. ISO/IEC 42001 provides certifiable lifecycle management, including bias audits, explainability requirements, and human oversight protocols. The NIST AI Risk Management Framework promotes structured risk identification and mitigation. The EU AI Act imposes legally binding obligations for high-risk systems, mandating human supervision and conformity assessments.

Despite differing approaches—voluntary standards, guidance frameworks, and binding regulation—these systems converge on three principles: explainability, human control, and auditability.

Geopolitically, competition over compute, energy, and semiconductor supply chains has intensified. The US–China semiconductor rivalry, European energy constraints, Gulf hyperscale investments, and India’s non-aligned governance strategy illustrate how infrastructure now defines power. Control over chips and data centres may matter more than control over territory.

6. Civil Liberties and Ethical Limits

While AI enhances security, it also threatens civil rights. Facial recognition, predictive policing, and mass surveillance risk normalising continuous monitoring. Empirical evidence shows algorithmic bias disproportionately affecting minority communities, leading to wrongful arrests and discriminatory enforcement.

The problem is not only technical but institutional. Proprietary algorithms obscure reasoning, limiting judicial review and democratic accountability. Function creep—where tools expand beyond their original purpose—gradually erodes civil liberties.

Governance frameworks therefore emphasise bias testing, transparency, and human oversight. Yet technical safeguards alone are insufficient. Public legitimacy requires participatory governance and legal redress mechanisms. Security systems that sacrifice dignity ultimately undermine the societies they aim to protect.

Human dignity must remain the limiting principle of AI deployment.

7. Leadership and Strategic Implications

Technological capability alone cannot resolve these tensions. Ethical leadership becomes decisive.

Effective intelligence leaders require strategic foresight, systems literacy, and ethical courage. They must understand not only how AI functions, but how it reshapes institutions, incentives, and power relations. This calls for what may be termed Mixed Reality Leadership: integrating human judgment with algorithmic speed.

Hybrid human–machine teaming—sometimes described as the “strategic centaur” model—combines computational analysis with human accountability. AI should augment rather than replace responsibility.

Without such leadership, governance frameworks risk becoming procedural compliance rather than meaningful oversight.

8. Conclusion: Toward Systems Governance

AI has transformed intelligence from discrete activity into continuous infrastructure. Open systems democratise analysis while covert systems weaponise perception. Cyber operations accelerate beyond human reaction times. Critical infrastructure, geopolitics, and civil liberties are increasingly intertwined.

Consequently, governance must evolve from regulating individual algorithms to governing entire socio-technical systems. This requires:

• transparency and auditability,

• meaningful human oversight,

• resilience and redundancy,

• and international norms to limit escalation.

The future of intelligence will not be decided solely by technical superiority, but by the ethical and institutional frameworks that shape its use. Societies that balance innovation with accountability will maintain both security and freedom. Those that fail risk either vulnerability or authoritarian control.

In the age of algorithmic intelligence, governance—not computation—becomes the decisive strategic advantage.