UPT Peace Assessment
Assessment Statement
Peace is already visible within the emerging artificial intelligence landscape, but primarily as an aspiration, ethical value, safety objective or beneficial social outcome rather than as a fully operationalised design condition.
Major international frameworks contain many of the conditions associated with peace: human dignity, accountability, transparency, fairness, sustainability, participation, human agency, resilience, international cooperation and concern for human and planetary wellbeing. The United Nations Educational, Scientific and Cultural Organization, which develops international standards across education, science, culture and ethics, goes particularly far: its global Recommendation on the Ethics of Artificial Intelligence calls for the peaceful use of AI and identifies living in “peaceful, just and interconnected societies” as a foundational value.[2] UNESCO
The report that provides the launch point for this assessment, Unlocking AI’s Potential to Serve Humanity, was produced through the United Nations University and the International Telecommunication Union. The United Nations University is the research and academic arm of the United Nations system, while the International Telecommunication Union is the United Nations specialised agency for digital technologies and global communications. The report explicitly identifies promoting peace alongside achieving the Sustainable Development Goals and responding to crises as one of the major challenges AI should help address.[1] AI for Good
The central gap identified by this assessment is therefore not an absence of concern for peace.
It is peace literacy.
Current frameworks are becoming increasingly sophisticated at recognising AI risk, trustworthy AI, responsible AI, human-rights impacts and ethical requirements. They are less developed at identifying what peace itself looks like within an AI system, distinguishing it from safety or absence of harm, recognising when fragmentation is transferred elsewhere in a system, and deliberately building peace-producing conditions into technology and the humans who create it.
Overall assessment: Peace capacity is emerging, but peace literacy remains underdeveloped.
Context
Artificial intelligence is moving rapidly from being a specialist technology into infrastructure affecting healthcare, education, communications, public administration, agriculture, environmental monitoring, employment, humanitarian response, national security and everyday human decision-making.
Unlocking AI’s Potential to Serve Humanity is significant because it does not treat beneficial AI as automatic. It identifies five conditions required for AI to serve human and planetary wellbeing: data quality, access and governance; digital infrastructure and access; AI literacy and talent; responsible AI policy; and digital ecosystem development. It concludes that beneficial AI development requires intentional, coordinated and multi-stakeholder action.[1] AI for Good
That creates an important peace-science proposition:
If AI for good must be intentional, AI for peace must also be intentional.
The question is whether the people and institutions developing AI currently possess sufficient peace literacy to know what they are trying to create.
There are strong foundations. The United Nations Educational, Scientific and Cultural Organization connects AI with dignity, inclusion, environmental wellbeing, autonomy, peaceful societies and human oversight. The Organisation for Economic Co-operation and Development, an intergovernmental organisation that develops policy standards across economic and social issues, promotes trustworthy AI based on human rights and democratic values. The U.S. National Institute of Standards and Technology, which develops technical standards and risk-management guidance, treats AI as inherently socio-technical rather than purely computational.[2][3][4] UNESCO
The pieces of peace are therefore present.
The question is whether they are forming a coherent architecture.
Peace Scan
3D — Physical Conditions
At the material level, AI already demonstrates considerable potential to support conditions in which peace becomes more possible. The United Nations University and International Telecommunication Union report examines practical applications across healthcare, food security, disaster response, disease monitoring, education, social assistance, biodiversity conservation, climate adaptation, energy optimisation and pollution reduction.[1] AI for Good
The United Nations Development Programme, the United Nations organisation focused on sustainable development and reducing poverty and inequality, has also explored human-in-the-loop AI for conflict anticipation and violence prevention. Its work demonstrates how machine learning can help humans process rapidly expanding crisis information while retaining human judgement in the decision-making process.[8] UNDP
These are genuine coherence capabilities. Health, food, infrastructure, environmental stability, access to information and early warning all affect whether societies remain stable under pressure.
However, the physical architecture also contains fragmentation.
AI requires computational infrastructure, communications networks, energy, data centres and access to high-quality data. The launch report identifies concentration of AI capability in a relatively small number of countries, substantial gaps in infrastructure, unequal representation within datasets and disparities in who can participate in AI development and governance.[1] AI for Good
A technology can therefore improve outcomes while simultaneously increasing dependency.
This is one of the first places where peace literacy changes the assessment.
The question is not merely:
Did the technology work?
It is also:
What happened elsewhere because it worked?
An AI system that raises agricultural productivity while increasing water consumption, farmer dependency, surveillance or economic concentration may create a measurable benefit without producing an equivalent increase in whole-system peace.
Peace requires the capacity to identify both effects.
4D — Historical Trajectory
The governance of artificial intelligence shows a familiar technological pattern: capability tends to develop more quickly than the institutions and social capacities needed to understand its wider consequences.
However, the current trajectory also shows substantial institutional learning.
The Organisation for Economic Co-operation and Development adopted its international AI Principles in 2019 and updated them in 2024. The United Nations Educational, Scientific and Cultural Organization adopted its global Recommendation on the Ethics of Artificial Intelligence in 2021. The U.S. National Institute of Standards and Technology released its Artificial Intelligence Risk Management Framework in 2023. The Council of Europe — the international organisation focused on human rights, democracy and the rule of law across Europe and partner states — opened the world’s first legally binding international AI treaty for signature in 2024.[2][3][4][5] UNESCO
The trajectory is therefore not one of regulatory absence. It is one of rapid learning following rapid technological development.
There is nevertheless a largely reactive sequence:
capability → consequences → recognised risks → governance response.
Peace literacy offers the possibility of another sequence:
understand peace conditions → design capability → monitor coherence → repair emerging fragmentation.
That is a significant distinction.
A mature peace architecture would not wait for visible harm before asking whether the system is losing coherence.
5D — Sovereignty, Identity and Relationships
This is one of the strongest areas of both opportunity and vulnerability.
AI changes not only what humans can do. It can change who gets to decide, who interprets information, who owns capability, whose identity is represented and who becomes dependent upon whom.
The launch report explicitly raises questions of community data ownership, unequal infrastructure, uneven participation in design and governance, digital divides and concentrated AI capability. It argues that accessibility and equity need to be built into infrastructure planning rather than treated as secondary consequences.[1] AI for Good
Other governance frameworks similarly recognise human agency. The Council of Europe’s Framework Convention requires AI activity to remain consistent with human rights, democracy and the rule of law throughout the AI lifecycle.[5] Council of Europe
The International Committee of the Red Cross — the independent humanitarian organisation mandated to protect people affected by armed conflict — provides an especially revealing stress test. Its work on AI in conflict argues that human control and judgement must remain meaningful where decisions have serious consequences for people’s lives, and warns against systems moving decision-making beyond meaningful human intervention.[7] International Review
This matters far beyond warfare.
A community may gain access to sophisticated agricultural technology while becoming dependent upon infrastructure controlled elsewhere.
A worker may become more productive while losing decision-making authority.
A citizen may receive more efficient government services while becoming less able to understand how decisions concerning them are made.
A person may gain access without gaining sovereignty.
Access and agency are not the same thing.
A peace-producing relationship between humans and AI should increase human capability without unnecessarily removing meaningful human choice, participation or responsibility.
6D — Competing Interpretations
The AI landscape contains remarkable agreement around terms such as responsible, safe, ethical, trustworthy, human-centred and beneficial.
There is much less agreement about exactly what these mean when values conflict.
The U.S. National Institute of Standards and Technology recognises that AI bias includes not only computational and statistical bias but systemic and human-cognitive bias. Human assumptions and expectations enter AI across design, deployment, operation and maintenance.[4] NIST Publications
This is particularly important for peace because “good” itself is interpretive.
A government may understand good AI partly through efficient service delivery.
A corporation may emphasise usefulness, innovation and adoption.
A human-rights organisation may prioritise autonomy and protection against discrimination.
A local community may understand benefit through livelihoods, cultural continuity and local control.
A security organisation may emphasise stability and threat prevention.
These perspectives can coexist. Fragmentation enters when one interpretation gains sufficient structural power that other affected actors can no longer meaningfully participate in defining the outcome.
AI also increasingly participates in interpretation itself. It classifies, ranks, predicts, recommends, generates explanations and influences what information people encounter.
That makes interpretive power part of the peace architecture.
Peace literacy therefore does not require eliminating disagreement.
It requires systems capable of holding competing interpretations without prematurely collapsing them into a single authoritative worldview.
Contestability, transparency, plural participation and meaningful human review become more than ethical safeguards.
They become peace-supporting conditions.
7D — Whole-System Behaviour
At the whole-system level, the emerging AI governance landscape shows meaningful movement toward coherence.
Human rights, sustainability, safety, inclusion, transparency, accountability, public participation, lifecycle management, international cooperation and human agency are increasingly being addressed together rather than as isolated concerns.
The U.S. National Institute of Standards and Technology describes AI as inherently socio-technical: its risks and benefits emerge through the interaction between technology, human behaviour, organisations and society.[4] NIST Publications
The United Nations University and International Telecommunication Union similarly conclude that AI for good requires a holistic ecosystem involving data, infrastructure, literacy, governance and relationships among government, business, academia and wider society.[1] AI for Good
This represents substantial coherence.
But the system remains incomplete.
Its strongest frameworks are better at operationalising responsibility, safety and risk than peace.
That matters because peace cannot be assumed simply because recognised harms have been reduced.
Within Peace As A Science, peace is treated as a stabilising configuration operating across physical conditions, time, relationships, meaning and whole-system coherence rather than simply as the absence of conflict.
The peace question for AI therefore becomes:
Does the AI system — together with the humans, organisations and institutions surrounding it — increase the capacity of the wider system to remain truthful, sovereign, relationally healthy, adaptive and coherent under pressure?
That is a different test from whether AI is merely useful or compliant.
Assessment of the 12 Peace Stabilisers
The Peace Stabiliser diagnosis reveals an uneven but promising landscape. The stabilisers are not equally visible in AI policy because some operate primarily through institutional and technical structures while others become more visible in the psychological and relational conditions of the humans designing those structures. The Peace Lab editorial standard specifically allows unequal emphasis while requiring the stabilisers to diagnose the condition of the system rather than simply describe the framework.
Honesty — Present but under pressure
AI governance gives substantial attention to transparency, explainability, traceability, data quality and accurate risk communication. This is a major coherence strength.
However, opacity, hallucinated information, poor-quality data, misinformation, inaccessible decision-making and uncertainty about system capabilities continue to destabilise signal integrity.
Honesty is recognised.
It is not yet consistently secured.
Open-mindedness — Emerging
Multi-stakeholder governance, multidisciplinary research and public consultation create important capacity for multiple perspectives to enter AI development.
Yet concentrated technological capability and unequal participation mean some perspectives have considerably more structural influence than others.
Peace requires open-mindedness not simply as tolerance of alternative opinions but as genuine system capacity to receive information that challenges existing assumptions.
Willingness — Emerging strongly at institutional level
The rapid growth of AI governance itself demonstrates willingness to adapt.
At the human level, however, willingness also means developers and institutions remaining receptive when evidence reveals that a system they have invested heavily in is generating unanticipated harm.
A psychologically defensive organisation may possess excellent technical expertise while still being difficult to correct.
Forgiveness — Low visibility, high psychological relevance
Forgiveness is largely absent from formal AI governance language, which appropriately emphasises accountability.
Its relevance appears more strongly inside development cultures.
Innovation inevitably involves mistakes. An environment in which admitting error produces humiliation or disproportionate punishment can incentivise concealment and defensiveness. An environment with no accountability can normalise avoidable harm.
The coherent position is responsibility with repair.
Forgiveness allows learning without erasing accountability.
Unity — Emerging but uneven
International cooperation is one of the strongest coherence indicators in the AI landscape. Standards organisations, governments, researchers, technology companies and multilateral institutions are increasingly communicating across sectors.
At the same time, infrastructure, computing resources, investment and decision-making influence remain unevenly distributed.
Global connectivity does not automatically create unity.
Unity requires participation without erasure of sovereignty.
Discipline — Relatively strong
This is one of the strongest current stabilisers.
Testing, evaluation, monitoring, lifecycle governance, risk assessment, auditing and continuous review are increasingly embedded within major governance frameworks.
The U.S. National Institute of Standards and Technology provides one of the clearest operational examples through four connected functions: Govern, Map, Measure and Manage.[4] NIST AI Resource Center
The important next question is what these disciplined processes are being trained to detect.
Faith — Low formal visibility, psychologically relevant
AI development operates under profound uncertainty.
The psychological risk is movement toward either uncritical technological optimism or catastrophic certainty.
Faith, in this context, is not a substitute for evidence. Its stabilising role is the human capacity to remain grounded in uncertainty without requiring false certainty, absolute control or despair.
That matters in teams responsible for technologies whose future consequences cannot be fully known.
Responsibility — Strong but incomplete
Responsibility is one of the clearest strengths in contemporary AI governance.
Accountability, impact assessment, oversight and organisational responsibility appear throughout international frameworks. The Council of Europe embeds accountability within its international legal architecture, while the United Nations Educational, Scientific and Cultural Organization calls for oversight, audit, due diligence and traceability throughout the AI lifecycle.[2][5] UNESCO
The challenge is distributed responsibility.
When datasets, foundation models, infrastructure providers, application developers, governments and end users are distributed across jurisdictions, responsibility can become diffused precisely when consequences become serious.
Love — Low formal visibility, foundational human relevance
AI governance rarely uses the word love.
It frequently uses functional relatives: dignity, care, wellbeing, inclusion and protection from harm.
Its strongest relevance may be psychological.
Developers can interact with millions of people they will never meet. At scale, there is a risk that humans become abstractions — users, datapoints, engagement statistics, risk categories or market segments.
Love acts as a counterweight to that abstraction by preserving the recognition that technological consequences are experienced by living people.
Wholeness — Present but inconsistent
Wholeness may be one of the most important stabilisers for this assessment.
Contemporary governance is increasingly moving toward socio-technical and lifecycle assessment. Yet individual sectors can still reward local success while overlooking transferred costs elsewhere.
An AI intervention can improve efficiency while increasing energy demand.
Improve agricultural yield while reducing farmer autonomy.
Improve security while increasing surveillance.
Improve personalised engagement while weakening human relationship.
A peace assessment asks whether the whole outcome remains coherent, not merely whether the target metric improved.
Joy — Low formal visibility, psychologically relevant
Joy is understandably not a common regulatory requirement.
But the culture in which AI is produced matters.
Teams operating under relentless competition, fear, urgency, burnout or existential pressure may become less psychologically flexible and less capable of reflective judgement.
Joy here does not mean forced positivity.
It indicates sufficient psychological spaciousness for curiosity, creativity, connection and the ability to imagine outcomes beyond threat management.
Peace — Explicit aspiration, incomplete operationalisation
Peace itself presents the clearest diagnostic result.
It is not absent.
The United Nations Educational, Scientific and Cultural Organization explicitly calls for peaceful AI systems and peaceful, just and interconnected societies. The United Nations University and International Telecommunication Union explicitly identify promoting peace as an objective for AI. The United Nations Development Programme is already exploring AI in violence prevention, while the United Nations Institute for Disarmament Research — the United Nations research institute specialising in disarmament and international security — examines AI’s implications for international peace and security.[1][2][6][8] UNESCO
Peace is therefore present in the landscape.
What remains weak is the ability to operationalise it.
The stabiliser diagnosis suggests that the AI landscape has developed substantial risk literacy, responsibility literacy and ethics literacy.
Its peace literacy is considerably less mature.
Peace Literacy: The Missing Capability
Peace literacy should not mean requiring programmers to adopt a particular political, spiritual or moral worldview.
It means developing the capacity to recognise the conditions that stabilise or fracture human systems.
For AI designers and decision-makers, that includes recognising when a technically successful system:
reduces human sovereignty;
concentrates interpretive power;
reproduces historical inequality;
weakens relationships;
creates dependency;
externalises environmental cost;
narrows meaningful choice;
or produces a benefit in one part of a system by destabilising another.
Current frameworks provide substantial foundations.
The United Nations Educational, Scientific and Cultural Organization already calls for AI literacy, data and information literacy, ethical training, critical thinking, teamwork, communication and socio-emotional skills. It also argues that public AI literacy is necessary for meaningful participation and protection from undue influence.[2] UNESCO
The United Nations University and International Telecommunication Union identify AI literacy and talent as one of their five pathways and note the value of combining computer science and engineering with social and environmental education.[1] AI for Good
This creates a natural opening.
Peace literacy does not need to compete with AI literacy. It can deepen it.
It asks those designing AI to understand not merely the tool, but the human system into which that tool is entering.
The Human Architecture Behind AI
One of the strongest findings of this assessment is that peace cannot be built entirely into code.
Artificial intelligence is created by humans operating within organisations, economies, cultures and relationships.
The U.S. National Institute of Standards and Technology explicitly recognises human-cognitive bias across the AI lifecycle and notes that organisational norms and decision-making structures can become embedded in technological systems.[4] NIST Publications
Peace literacy extends this observation.
Programmers, researchers, executives and regulators bring assumptions about competition, control, productivity, uncertainty, human nature, responsibility and progress into the systems they create.
This is where the less visible Peace Stabilisers become highly relevant.
Can a team admit error?
Can it listen when affected communities disagree?
Can developers tolerate uncertainty?
Can disagreement occur without relational rupture?
Can responsibility be accepted without scapegoating?
Can people retain empathy for populations represented only through data?
Can teams remain creative rather than perpetually threat-driven?
These may appear to be psychological questions.
They are also technological questions because the humans answering them are building the technology.
The peace capacity of AI cannot be separated entirely from the peace capacity of the people designing it.
Identifying Peace in AI
This assessment does not find a single metric capable of determining whether AI is peaceful.
Peace appears instead as a pattern across multiple conditions.
An AI system moves toward peace when:
its information remains sufficiently truthful to support reliable decisions;
people retain meaningful agency and contestability;
relationships become more capable rather than unnecessarily dependent;
power does not become invisible simply because it is computational;
different perspectives can remain present;
benefits are not systematically achieved by transferring harm elsewhere;
human responsibility remains identifiable;
environmental consequences are incorporated;
errors can be detected and repaired;
and the wider system becomes more capable of remaining coherent under pressure.
This allows us to distinguish several concepts that are often placed close together.
Safe AI asks whether unacceptable harms are prevented.
Responsible AI asks whether appropriate duties and processes are being upheld.
Trustworthy AI asks whether the system possesses qualities that justify reliance.
Peace-supporting AI asks whether the wider human and planetary system becomes more capable of sustaining coherent relationships and outcomes.
These categories overlap.
They are not interchangeable.
Building Peace into AI
The intervention suggested by this assessment is not simply adding “peace” as another principle beside safety, fairness, transparency and accountability.
That could leave the underlying architecture unchanged.
Peace becomes more useful when it operates across the AI lifecycle.
At the beginning, peace literacy helps determine how a problem is framed.
During design, it helps identify sovereignty, relationships and possible transferred costs.
During testing, it broadens impact assessment beyond immediate performance.
During deployment, it asks whether affected people retain meaningful agency.
During monitoring, it looks for emerging fragmentation that conventional performance metrics may miss.
During failure, it supports accountability, repair and learning.
This approach does not require abandoning existing governance architecture.
It could extend it.
For example, the U.S. National Institute of Standards and Technology already asks organisations to Govern, Map, Measure and Manage AI risk.[4] NIST AI Resource Center
A peace-oriented extension asks:
Can we also govern, map, measure and manage the conditions that allow peace to hold?
That may be one of the most practical openings identified by this assessment.
Intervention Points
The highest-leverage intervention is peace literacy across the AI ecosystem.
That includes programmers and engineers, but should not be limited to them. Product designers, executives, investors, educators, policymakers, regulators, researchers and organisations deploying AI all influence what eventually reaches society.
A second leverage point is peace assessment within the AI lifecycle. Existing impact assessments could be supplemented by questions about sovereignty, relational consequences, historical patterns, competing interpretations and transferred fragmentation.
A third is whole-system evaluation. AI success should not be determined solely by whether it solves the problem it was instructed to solve. Evaluation should also examine whether it destabilises another part of the human or planetary system.
A fourth is human development within technical cultures. The psychological Peace Stabilisers suggest that accountability alone is insufficient. Teams also require the capacity for openness, willingness, healthy repair, uncertainty tolerance, care, integration and creativity.
The goal is not to make programmers peace activists.
It is to give those shaping powerful systems greater capacity to recognise the difference between technical optimisation and systemic coherence.
Future Outlook
If current dynamics continue without a clearer peace framework, AI governance is likely to become increasingly sophisticated at identifying recognised categories of harm while remaining less capable of seeing fragmentation that falls between them.
AI may become safer, more accountable, more transparent and more rights-aware while still producing forms of dependency, concentrated interpretive power, relational deterioration or ecological displacement that no single existing assessment category captures.
That is not a failure of current frameworks.
It indicates the next layer of development.
An alternative trajectory is already possible because many of the required components exist.
International governance is becoming more systemic. Human agency is receiving greater attention. Environmental consequences are entering AI policy. AI literacy is expanding beyond technical competence. Multi-stakeholder governance is growing. International institutions are beginning to speak explicitly about AI and peace.[1][2][5][6] AI for Good
Peace literacy could connect these developments.
If that occurs, the central AI question may gradually shift from:
How do we prevent AI from causing harm?
toward:
What conditions should AI help humanity stabilise?
That is a much larger ambition.
It is also increasingly possible.
Peace Assessment
The contemporary AI landscape demonstrates significant emerging coherence but incomplete peace architecture.
Its greatest strength is that leading international organisations no longer treat AI as a purely technical system. Human rights, sustainability, participation, social consequences, accountability, human judgement and international cooperation are increasingly recognised as integral to AI governance.
Its greatest weakness is not the absence of peace-related values.
It is their fragmentation.
The AI landscape already contains many of the components from which peace can be built. What is largely missing is an integrated capacity to recognise these components as conditions of peace, detect when they begin to fracture, and intentionally design them into technological and human systems.
This makes peace literacy the principal leverage point emerging from the assessment.
UPT Summary
The global AI landscape contains substantial coherence potential. Leading frameworks increasingly connect artificial intelligence with human dignity, accountability, safety, fairness, sustainability, participation, human agency and international cooperation. Most significantly, prominent United Nations frameworks now explicitly identify peaceful societies and the promotion of peace as legitimate objectives of AI development.
The primary fragmentation is therefore not a lack of ethical concern but a lack of integration. Current frameworks can increasingly identify unsafe, discriminatory, opaque or irresponsible AI, yet they remain less equipped to determine whether an apparently beneficial intervention strengthens sovereignty and relationships, transfers harm elsewhere, reproduces historical fragmentation or improves whole-system coherence.
The overall peace assessment is emerging but incomplete. Responsibility and discipline appear comparatively strong. Honesty, open-mindedness, unity and wholeness are present but uneven. Willingness, forgiveness, faith, love and joy become especially significant within the psychological and relational environments of the humans creating AI. Peace itself is increasingly named, but remains stronger as an aspiration than as an operational design condition.
The most important leverage point is peace literacy: enabling those who design, govern and deploy AI to recognise the conditions that stabilise or fracture human and planetary systems.
If current dynamics remain unchanged, AI governance will continue improving its capacity to manage recognised risks while some forms of systemic fragmentation remain difficult to detect. If peace literacy and peace assessment are integrated into existing AI design and governance processes, artificial intelligence could move beyond being merely safer or more responsible toward intentionally strengthening human and planetary coherence.
The significance of this case is not that AI needs another ethical rule. It is that humanity increasingly needs the capacity to recognise peace with enough clarity to build for it.
About the UPT Peace Assessment
A UPT Peace Assessment applies Unified Peace Theory as a practical analytical framework to examine the health, resilience and likely trajectory of a human system. Its purpose is to identify the conditions that strengthen or weaken long-term peace, stability and coherence, while keeping the focus on the issue being assessed rather than on the methodology itself.
Endnotes and References
[1] United Nations University & International Telecommunication Union. Unlocking AI’s Potential to Serve Humanity: Robotics, Geospatial AI and Communications Networks. This report is the primary launch point for the assessment and examines practical AI applications, human and planetary wellbeing, five enabling pathways for AI for Good, and the explicit objective of promoting peace. Read the report. AI for Good
[2] United Nations Educational, Scientific and Cultural Organization. Recommendation on the Ethics of Artificial Intelligence, 2021. UNESCO’s global normative framework addresses human dignity, human rights, sustainability, AI literacy, peaceful and interconnected societies, human oversight and ethical governance throughout the AI lifecycle. Read the Recommendation. UNESCO
[3] Organisation for Economic Co-operation and Development. OECD AI Principles, adopted 2019 and updated 2024. The principles provide an international policy baseline for innovative and trustworthy AI that respects human rights and democratic values. Read the AI Principles. OECD
[4] U.S. National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework (AI RMF 1.0), 2023. NIST provides an operational framework for trustworthy and responsible AI and treats AI as a socio-technical system shaped by technical, organisational and human factors. Read the framework. NIST Publications
[5] Council of Europe. Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, opened for signature in 2024. The Convention is the first legally binding international treaty specifically addressing AI and requires AI lifecycle activities to remain consistent with human rights, democracy and rule of law. Read about the Convention. Council of Europe
[6] United Nations Institute for Disarmament Research. Artificial Intelligence in the Military Domain and Its Implications for International Peace and Security: An Evidence-Based Road Map for Future Policy Action, 2025. The institute examines the implications of military AI beyond autonomous weapons and places AI explicitly within international peace-and-security analysis. Read the report. UNIDIR → Building a more secure world.
[7] International Committee of the Red Cross. Artificial Intelligence and Machine Learning in Armed Conflict: A Human-Centred Approach. The ICRC examines AI through humanitarian law, civilian protection, human agency and meaningful human control in high-consequence environments. Read the paper. International Review
[8] United Nations Development Programme. Harnessing the Potential of Human-in-the-Loop Artificial Intelligence for Risk Anticipation and Violence Prevention, 2023. The paper explores practical applications of AI for conflict analysis, crisis anticipation and violence prevention while retaining human participation in decision-making. Read the paper. UNDP
[9] United Nations Advisory Body on Artificial Intelligence. Governing AI for Humanity: Final Report, 2024. The report proposes an inclusive and distributed global architecture for AI governance based on international cooperation and the protection of human rights. Read the report. Digital Library
Framework source: Michelle Lightworker, Peace As A Science: Towards a Unified Peace Theory, particularly the dimensional ecology of peace and the Peace Stabilisers used as the analytical and diagnostic foundations of this assessment.
Editorial methodology: UPT Peace Lab Assessment Editorial Standards v2.0. The standards distinguish the 3D–7D Peace Scan as the investigative phase and the 12 Peace Stabilisers as the diagnostic phase, while requiring the assessment to remain focused on the issue rather than the methodology.