SITA: Lessons in Tech Governance

‍SITA (Société Internationale de Télécommunications Aéronautiques) represents a remarkable 75-year experiment in international technology cooperation through industry-owned infrastructure. Founded in February 1949 by 11 airlines as the Cold War began, SITA grew from a modest telecommunications network into the world's largest packet-switched data network by 1981, serving 340 airlines across 200+ countries. Today, it remains a 100% industry-owned cooperative generating $1.5 billion in annual revenue while serving 90% of global airline business. However, a critical finding emerges: SITA's Cold War success operated primarily within the Western alliance rather than truly bridging the Iron Curtain, as the Soviet Union never joined ICAO or IATA during this period. This nuance makes SITA's lessons for AI governance both instructive and cautionary.

The cooperative that built the internet's precursor

SITA's founding in February 1949 addressed a simple economic problem: individual airlines couldn't afford separate global telecommunications networks. Eleven carriers—British European Airways, British Overseas Airways, British South American Airways, KLM, Sabena, Swissair, TWA, Swedish Aerotransport, Danish Air Lines, Norwegian Air Lines, and Air France—pooled resources to create shared infrastructure. The first telecommunications center opened in Rome in 1949, connecting Paris, Nice, and Frankfurt via manual teletype systems using perforated tape.

What distinguished SITA was its cooperative ownership structure. Airlines owned and governed SITA collectively, making them simultaneously users and shareholders. This aligned incentives: members didn't want accidents or failures affecting their own operations, creating natural pressure toward safety and reliability over profit maximization. The cooperative operated on a cost-recovery basis, reinvesting revenues into infrastructure improvements benefiting all members equally.

The network's technical evolution proved prescient. By 1966, SITA launched the world's first working packet-switching network—three years before ARPANET. In 1969, SITA activated the first nodal distributed network, contributing foundational concepts to internet development. By 1981, SITA deployed its third-generation Data Transport Network using X.25 protocol, becoming the world's largest public data network with 120,000 terminals across 187 countries processing 36 billion transactions annually.

Governance designed for political neutrality

SITA's governance architecture deliberately insulated it from Cold War political pressures through multiple mechanisms. The headquarters location in Geneva, Switzerland—a permanently neutral nation since 1815—provided crucial diplomatic cover. As an industry-owned cooperative rather than government-controlled enterprise, SITA operated one level removed from direct state interference.

The General Assembly served as ultimate authority, with member airlines voting on major strategic decisions. A Board of Directors, elected by members for renewable three-year terms, supervised operations. Geographic representation ensured no single political bloc dominated decision-making. In 2021, governance reforms streamlined the structure to a single smaller Board plus a 20-member SITA Council representing 10 geographic groups, maintaining the principle of distributed power.

Technical standards development followed a multi-stakeholder consensus process. SITA participates in approximately 55 different Standards Setting Working Groups with 40 dedicated participants, collaborating with IATA, ICAO (International Civil Aviation Organization), ITU (International Telecommunication Union), FAA, and other bodies. Standards emerged from operational needs rather than political considerations. The cooperative structure meant decisions required broad member support—no single airline held veto power.

This neutral technical mandate proved essential. SITA's articles of association stated its objective: "to foster all telecommunications and information processing required in the operation of the air transport industry with the aim of promoting in all countries safe and regular air transport." Universal service transcended political boundaries. The network complemented ICAO's Aeronautical Fixed Telecommunication Network, helping airlines fulfill Chicago Convention obligations—the 1944 international treaty establishing aviation cooperation principles.

Technical interoperability as political necessity

SITA's technical services created unavoidable interdependence among airlines, compelling cooperation even amid geopolitical tensions. The Type B messaging system, developed in the early 1960s, established universal standards for aviation communications. With a simple structure of 60 lines × 63 characters using capitalized letters, numbers, and three symbols from teletype character sets, Type B messages enabled flight plans, passenger bookings, baggage tracking, and operational coordination. By 2019, SITA processed approximately 188 million Type B messages daily.

The 7-character IATA addressing system provided politically neutral routing: three characters for airport/location code, two for office/department, two for airline designator. This standardization meant a message from New York to Paris routed identically regardless of whether sent by American Airlines or Air France. The store-and-forward architecture with guaranteed delivery ensured messages crossed borders despite intermittent connectivity—critical when telecommunications links traversed politically sensitive boundaries.

SITA's X.25 packet-switching network adoption in 1981 leveraged international standards developed through ITU-T, a UN specialized agency. This multilateral standardization process included participation from both Western and Soviet telecommunications authorities, creating common technical language. X.25's three-layer architecture (physical, data-link, and network layers) functioned identically across political systems. Permanent Virtual Circuits (PVCs) and Switched Virtual Circuits (SVCs) enabled flexible connectivity while maintaining security through logical channel identifiers.

Beyond messaging, SITA provided critical reservation system connectivity through AIRIMP (Airline Industry Reservations Interline Message Procedures) standards, enabling millions of daily transactions between airlines and travel agencies. WorldTracer, developed jointly with IATA in the early 1990s, created the global baggage tracking system now used by 500+ airlines and 2,800+ airports. ACARS (Aircraft Communications Addressing and Reporting System), deployed in 1978, established air-to-ground data links for flight operations. Each system embedded interoperability requirements that made unilateral action by individual airlines operationally impossible.

The Cold War reality: Western success with Eastern limitations

The research reveals a critical nuance often overlooked in accounts of SITA's Cold War operations: Soviet participation was limited or potentially non-existent during the Cold War era. A 1964 academic paper notes explicitly: "The USSR is one of the few major governments that does not belong to the International Civil Aviation Organization (ICAO) nor does Aeroflot belong to the International Air Transport Association (IATA)."

This fundamentally reframes SITA's Cold War narrative. Rather than demonstrating successful East-West technology cooperation across the Iron Curtain, SITA succeeded by operating primarily within the Western alliance and non-aligned countries. Expansion from 52 members (1957) to 340 airlines (1980s) represented growth mainly across NATO, neutral nations, and developing countries aligned with the West—not deep integration with COMECON (Council for Mutual Economic Assistance) airlines behind the Iron Curtain.

Evidence of East-West aviation cooperation did exist, but remained limited. The 1968 Moscow-New York route between Aeroflot and Pan Am symbolized détente but represented isolated bilateral agreements rather than systematic technical integration. COMECON airlines—Polish LOT, Czechoslovak CSA, Hungarian Malév, Bulgarian Balkan, Romanian TAROM, East German Interflug—operated entirely Soviet-built aircraft fleets and likely developed parallel telecommunications systems.

Technology export restrictions reinforced this divide. CoCom (Coordinating Committee for Multilateral Export Controls) classified telecommunications equipment and computer technology as strategic, limiting Western technology transfer to the Soviet bloc. While processes of technological standardization did facilitate some East-West knowledge transfer, comprehensive integration awaited the Cold War's end.

Modern confirmation exists that Aeroflot and Eastern European airlines now participate in SITA (Aeroflot deployed SITA CrewTablet in 2014 and serves on the SITA Council), but specific documentation of when and how Soviet carriers participated during 1949-1991 remains absent from accessible sources. This information gap likely reflects operational security concerns, limited archival access, and the reality that comprehensive cooperation simply didn't occur until after 1991.

Why SITA succeeded within its sphere

Despite geographical limitations, SITA achieved remarkable success within its operational domain through several converging factors. Economic incentives proved overwhelming. Airlines faced telecommunications costs they couldn't individually afford. Shared infrastructure reduced per-airline investment by orders of magnitude while providing superior global connectivity. The cooperative model meant cost savings benefited all members rather than enriching external shareholders.

Safety imperatives transcended commercial competition. Aviation's life-and-death stakes created absolute requirements for reliable communications. Weather data sharing, air traffic control coordination, emergency response, and maintenance reporting couldn't fail without catastrophic consequences. ICAO documentation explicitly notes: "By meeting Member airlines worldwide telecommunication needs, SITA enables them to fulfill their responsibilities under the Annexes to the Chicago Convention on International Civil Aviation." Safety requirements compelled cooperation even among fierce commercial rivals.

Network effects created powerful lock-in. Each additional airline joining SITA exponentially increased the network's value to all existing members. By 1989, SITA served 460 airlines across 31,000 offices—achieving near-universal coverage made membership effectively mandatory for international carriers. Opting out meant operational isolation. This dynamic explains rapid expansion once critical mass was achieved.

The Chicago Convention framework provided essential legal infrastructure. Signed in 1944 by 52 states, the Convention established sovereignty principles, international safety standards, and ICAO as coordinating body. This created normative expectations for aviation cooperation that SITA operationalized. Airlines could justify technical cooperation to skeptical governments by referencing treaty obligations.

Industry ownership insulated SITA from direct political control. Unlike government-run telecommunications monopolies subject to political interference, SITA answered to airline members focused on operational efficiency. This commercial mandate provided cover for maintaining cooperation during diplomatic tensions. When Western governments pressured airlines to reduce Soviet engagement, airlines could cite operational necessity.

Comparing SITA to other Cold War cooperation models

SITA's model shares similarities with other successful Cold War international technical organizations while exhibiting distinctive characteristics. CERN (European Organization for Nuclear Research), founded in 1954, brought together former wartime enemies for fundamental physics research. CERN explicitly "embodied the ideal of scientific internationalism" and maintained cooperation with the Soviet Union, even becoming "the first western institute to sign agreements with Soviet scientific institutes" at the Cold War's height. Yugoslavia's membership bridged East-West divides.

However, critical differences emerge. CERN required government appropriations and functioned as an intergovernmental organization subject to political oversight. SITA, self-funding through member fees and industry-owned, avoided this vulnerability. CERN conducted long-term research projects; SITA enabled daily commercial operations. Airlines couldn't afford to let political tensions disrupt routine flights, creating stronger continuity pressures than scientific collaboration.

ITU (International Telecommunication Union), founded in 1865 and later becoming a UN specialized agency, provided telecommunications coordination with near-universal membership (194 countries). ITU's intergovernmental structure made it politically neutral but also subject to diplomatic pressures. SITA's industry ownership and aviation-specific focus enabled operation one level removed from government politics. ITU primarily set standards; SITA actually operated networks. This distinction between standardization and operational implementation proved significant.

What made SITA unique was its combination of: cooperative ownership by commercial entities rather than states; operational criticality requiring daily functionality; technical focus on narrow practical problems; economic model enabling self-sustaining investment; and deep specialization creating expertise concentration. This combination created unusually strong incentives for continued cooperation even when political tensions might otherwise disrupt collaboration.

SITA's contemporary relevance and scale

SITA's cooperative model not only survived but thrived into the 21st century. The 2023 annual report confirms SITA remains "100% owned by the industry and driven by its needs." Revenue reached $1.487 billion (up 7% from 2022), with 5,100 employees from 130+ nationalities serving 2,500 customers including 400 member airlines, 1,000+ airports, 18,000+ aircraft, and 70+ governments across 200+ countries and territories.

The governance structure evolved while preserving cooperative principles. Following extensive 2014-2015 reforms, SITA streamlined from a two-tiered board to a single 12-member Board of Directors plus the 20-member SITA Council representing 10 geographic groups. This modernization aimed to "align with global best practices for commercial cooperatives" while maintaining member voice through the Council's advisory role on portfolio strategy, membership matters, and long-term direction.

SITA's service portfolio demonstrates continued technical leadership. Five business lines generated 2023 revenues: SITA at Airports ($615M) for passenger processing and baggage management; ATI Communications & Data Exchange ($391M) for networking and messaging; SITA for Aircraft ($296M) for flight operations; SITA at Borders ($132M) for border management; and CHAMP Cargosystems ($57M) for cargo operations. The company holds 377 granted patents and continues investing in cloud transitions, biometrics, AI integration, and sustainability solutions.

Critically, the cooperative model remains functionally intact despite technology and market evolution. Members own certificates representing stakes in SITA SC (parent cooperative), which holds 45.23% of SITA N.V. directly with the SITA Group Foundation holding 54.77% on behalf of certificate holders. Recent reforms introduced dividend payments while maintaining the core principle: revenues serve member interests rather than external shareholders. This 75-year continuity demonstrates cooperative governance can scale to global operations generating billions in revenue while maintaining member alignment.

Lessons for AI governance: Transferable insights

Multiple experts identify aviation, and SITA specifically, as the most relevant model for AI governance. Dr. Stuart Russell of UC Berkeley, speaking at the 2024 AI for Good Summit, stated: "Air travel, once fraught with risk, is now one of the safest modes of transport thanks to rigorous regulatory standards." He specifically identified aviation as the best industry model for AI regulation.

A comprehensive 2020 Berkeley CLTC study titled "Flight to Safety-Critical AI" examined aviation's approach to AI adoption, finding: "Industry experts report that representatives from firms, regulatory bodies, and academia have engaged in a highly collaborative AI standard-setting process, focused on meeting rather than relaxing aviation's high and rising safety standards." The research identified "limited evidence of an AI 'race to the bottom' and some evidence of a (long, slow) race to the top," demonstrating that industry cooperation on safety need not sacrifice innovation velocity.

Several specific parallels suggest strong transferability from SITA's model to AI governance:

Technical interoperability requirements mirror aviation's challenges. Just as airlines needed standardized messaging protocols for safety-critical operations across borders, AI systems require interoperability standards for model interfaces, data exchange formats, common evaluation frameworks, and standardized APIs for cross-system integration. SITA's success developing Type B messaging, X.25 protocols, and AIRIMP reservation standards through multi-stakeholder working groups offers proven mechanisms for achieving technical consensus across competitive entities.

Shared infrastructure for safety-critical functions represents high-transferability opportunities. SITA's cooperative model enabled smaller airlines to access telecommunications capabilities matching larger carriers. An AI analogue could provide shared safety testing infrastructure—red-teaming capabilities, evaluation frameworks, benchmark datasets, and auditing tools—where expensive safety research costs are distributed across participants. This addresses current concerns that comprehensive AI safety testing may be prohibitively expensive for all but the largest firms.

Multi-layer governance architecture from aviation translates effectively. The structure proceeds: International Treaty Level (Chicago Convention) → Government Coordination (ICAO) → Industry Cooperative (SITA) → Individual Company Implementation. AI governance could similarly layer: international AI safety principles → government coordination body → industry cooperative for technical standards → individual firm deployment. This distributes functions appropriately: governments set safety requirements, industry cooperatives develop implementation standards, and individual companies innovate within frameworks.

Cross-border cooperation mechanisms demonstrate feasibility despite geopolitical tensions. While SITA's Cold War East-West cooperation was more limited than sometimes portrayed, it successfully maintained operations across 200+ countries through politically turbulent decades. Today, SITA continues operating across diverse political systems, handling sanctions compliance (terminating Russian operations in 2022) while maintaining global network functionality. For AI governance facing U.S.-China tensions and regulatory fragmentation across EU, U.S., China, and the Global South, SITA's neutral, technically-focused, industry-owned structure offers a proven pathway.

Incremental deployment with safety gates directly applies to AI. Aviation's approach—extensive testing before deployment, phased rollout strategies, continuous monitoring, failure mode analysis, and mandatory incident reporting—created the sector's exceptional safety record. SITA's innovation track record (first packet-switching network in 1960s, common use terminals in 1980s, biometric processing in 2000s, AI optimization in 2020s) demonstrates that rigorous safety requirements and rapid innovation can coexist when properly structured.

Critical limitations and non-transferable aspects

However, significant differences between aviation and AI limit direct applicability of SITA's model in several areas.

Clear safety metrics distinguish aviation from AI. Aviation measures success through unambiguous outcomes: accident rates, near-misses, equipment failures, on-time performance. Decades of incident data enable actuarial analysis and evidence-based regulation. AI lacks comparable clarity. Defining "AI safety" or "alignment" remains contested. Many AI harms—algorithmic bias, privacy violations, economic displacement, misinformation amplification—are diffuse, delayed, or subjectively assessed. This measurement challenge complicates creating SITA-equivalent safety standards for AI.

Industry maturity and consolidation also differ sharply. Aviation features a relatively stable set of major airlines, aircraft manufacturers, and equipment suppliers with high barriers to entry (certifying new aircraft costs billions and takes years). AI's rapid evolution sees constant new entrants, with lower barriers to model development and deployment. This dynamism challenges cooperative governance requiring stable membership and long-term commitments.

Physical infrastructure localization that grounded aviation in specific jurisdictions has no AI equivalent. SITA's telecommunications centers, airports, and aircraft physically existed within national territories subject to local regulation. Digital AI systems deploy globally instantly, reducing compliance pressure from physical presence. While data centers and chip fabrication plants create some localization, the virtual nature of AI models fundamentally differs from aviation's physical constraints.

Mission criticality consensus exists for aviation—plane crashes are universally unacceptable—but AI lacks equivalent agreement. Substantial disagreement persists about AI risk severity, timelines, and appropriate interventions. Aviation achieved consensus through visible disasters creating public demand for regulation. AI's impacts, while potentially profound, often manifest gradually or invisibly. Building comparable consensus faces greater challenges.

Dual-use concerns affect AI more acutely than civil aviation. While aviation technology has military applications, commercial aviation maintains clear civilian character. AI's military, surveillance, and information control applications blur civilian-military boundaries. This complicates international cooperation, as nations view AI development through strategic competition lenses more intensely than civil aviation.

Structural recommendations for AI cooperative governance

Drawing on SITA's 75-year experience while accounting for AI's distinct characteristics suggests several architectural principles for AI governance.

Create an AI Industry Cooperative for Safety Infrastructure modeled on SITA's cooperative structure. This entity would be industry-owned, non-profit, with global reach, governed by member-elected boards with expert advisory councils. Primary purpose would be shared AI safety testing, evaluation frameworks, and benchmark datasets. Membership would include AI developers, deployers, researchers, and civil society representatives. Funding through member dues scaled to organizational capability would be reinvested in safety research and infrastructure.

Establish Multi-Layer Governance Architecture separating functions appropriately. An international treaty level would establish high-level principles (analogous to Chicago Convention). Government coordination bodies would develop binding standards (analogous to ICAO). Industry cooperatives would create technical implementation standards (analogous to SITA). Individual companies would innovate within these frameworks. This separation prevents governments from micromanaging technical details while ensuring industry standards serve public safety rather than only commercial interests.

Develop Sector-Specific Implementation recognizing AI's diversity. Different use cases—healthcare AI, autonomous vehicles, financial systems, content recommendation, scientific research—require tailored governance analogous to SITA's business lines. Shared safety infrastructure supports specialized deployment standards, avoiding one-size-fits-all approaches that either under-regulate high-risk applications or over-regulate low-risk uses.

Create Graduated Standards Framework with risk-based tiers. Level 1 establishes basic safety requirements for all AI systems. Level 2 imposes enhanced requirements for high-risk applications. Level 3 applies frontier model requirements for most capable systems. This parallels aviation's certification levels (light aircraft, commercial airliners, experimental aircraft) that calibrate oversight to risk.

Establish Neutral Convening Platform for regular coordination. Annual AI Safety Assemblies (analogous to SITA's Annual General Meeting) would bring together stakeholders for standards updates, incident reviews, and coordination. Regular working groups on specific technical challenges would enable pre-competitive collaboration on shared problems. This creates relationship infrastructure enabling coordination during crises.

Implementation considerations and realistic constraints

Operationalizing SITA-inspired AI governance faces significant practical challenges requiring clear-eyed assessment. Power concentration in AI exceeds aviation's. A handful of companies—OpenAI, Google DeepMind, Anthropic, Meta, and a few others—control frontier model development. This oligopoly structure creates different dynamics than aviation's hundreds of airlines. Cooperative governance requires these frontier developers' participation, giving them substantial influence over standards they'll be judged against. Designing governance preventing regulatory capture while maintaining cooperation demands careful attention to voting structures, transparency requirements, and external oversight.

Fragmented AI ecosystem far exceeds aviation's complexity. Beyond frontier labs, thousands of companies fine-tune models, develop applications, create tools, and deploy AI across innumerable contexts. Aviation's value chain—aircraft manufacturers, airlines, airports, ATC—is comparatively simple. AI governance must accommodate this diversity without creating compliance burdens that stifle beneficial innovation. Tiered approaches distinguishing frontier development, application deployment, and end-use may prove necessary.

Faster innovation pace in AI challenges standards development timelines. Aviation technology evolves over decades; new aircraft families take 7-10 years from design to service. AI capabilities advance monthly. SITA's standards development—working groups, consensus building, testing, gradual rollout—takes years. AI governance requires more agile processes while maintaining rigor. Provisional standards with sunset clauses, continuous review cycles, and adaptive frameworks may be necessary.

International coordination complexity in AI's multipolar world differs from aviation's Cold War context. While SITA operated primarily within the Western alliance with limited Soviet participation, AI governance must somehow accommodate U.S.-China tensions alongside EU regulatory leadership and Global South participation. Aviation benefited from U.S.-European shared democratic values and security alliance. AI faces competing governance philosophies: EU ethics-first regulation (AI Act), U.S. innovation-first approach, China's national alignment mandate, and Global South concerns about access and exploitation. Creating governance bridging these divides may prove more difficult than SITA's already-challenging Cold War environment.

Liability and accountability frameworks remain unresolved for AI unlike aviation's mature structures. Aviation established clear liability: airlines for passenger safety, manufacturers for equipment defects, pilots for operational decisions. AI's distributed development (base model creators, fine-tuners, deployers, users) obscures accountability. Who bears liability when a fine-tuned model produces harmful outputs? Open-source model creators? Deployment platforms? End users? Without resolving these questions, cooperative governance lacks enforcement mechanisms.

The SITA model's enduring insights

Despite limitations and complications, SITA's 75-year track record offers several enduring insights for technology governance under geopolitical stress. Most fundamentally, cooperative ownership can align safety and innovation incentives. When participants share both financial risks and operational consequences, they naturally prioritize reliability. SITA's members don't want telecommunication failures affecting their flights, creating organic safety focus without external enforcement. AI developers participating in cooperative safety infrastructure would similarly internalize consequences of inadequate testing or evaluation.

Industry-led standards development works when properly embedded in broader governance ecosystems. SITA doesn't operate in isolation—it functions within ICAO governmental coordination, IATA industry association frameworks, national regulatory oversight, and international treaty obligations. This multi-layer architecture provides checks and balances. Similarly effective AI governance likely requires industry cooperatives for technical standards development situated within governmental oversight, international coordination, and civil society accountability. Neither pure industry self-regulation nor pure governmental control will suffice.

Shared infrastructure reduces costs while improving quality through economies of scale. SITA enables small airlines to access telecommunications infrastructure matching major carriers' capabilities. AI safety testing, particularly for frontier models, may become prohibitively expensive for individual companies. Red-teaming, adversarial testing, extensive evaluation across diverse scenarios, and long-term monitoring could cost hundreds of millions or billions of dollars. Cooperative infrastructure spreads costs while ensuring even smaller AI developers can conduct adequate safety testing.

Technical necessity creates cooperation opportunities transcending political tensions. Airlines couldn't function internationally without interoperability, compelling cooperation even between geopolitical rivals. AI may develop similar imperatives. As AI systems become embedded in critical infrastructure—healthcare, energy grids, financial systems, telecommunications, transportation—interoperability and safety standards may become unavoidable requirements. This technical necessity could provide leverage for cooperation even amid U.S.-China strategic competition, much as aviation safety standards persist despite geopolitical tensions.

Multi-decade evolution requires patience and persistence. SITA took decades to mature from an 11-airline European network to global infrastructure serving 90% of world airline business. Early years involved experimentation, failures, and incremental expansion. AI governance advocates should expect similar trajectories. Calls for comprehensive global AI governance agreements may be premature. Incremental institution-building—establishing working groups, developing narrow technical standards, creating shared testing infrastructure, building trust through small collaborations—may prove more realistic than grand treaties.

A model of bounded but real success

SITA represents neither a perfect template for AI governance nor an irrelevant historical curiosity. The research reveals a more nuanced reality: SITA succeeded remarkably within its operational domain—primarily Western alliance and non-aligned countries—while facing significant limitations in achieving true East-West integration during the Cold War. Soviet non-participation in ICAO and IATA, and unclear SITA membership during 1949-1991, demonstrates that even successful international technical cooperation hits limits when geopolitical divides run deep.

This bounded success offers realistic rather than utopian lessons for AI governance. Technology cooperation across political boundaries is achievable but difficult. Industry-owned cooperatives can maintain technical collaboration when governments clash. Economic incentives and safety imperatives can sustain coordination despite diplomatic tensions. But comprehensive global integration may require geopolitical conditions—shared values, mutual trust, aligned incentives—that don't always exist. SITA thrived within the Western alliance precisely because democratic governments, market economies, and security partnerships created favorable conditions. East-West integration awaited the Cold War's end.

For AI governance, this suggests both opportunities and constraints. Within the U.S.-European-allied democratic sphere, SITA-inspired cooperative governance appears feasible. Shared democratic values, regulatory coordination, security partnerships, and economic integration create conditions analogous to SITA's Cold War environment. Extending such governance to incorporate China—the world's second major AI power—faces steeper challenges absent political conditions enabling trust and reciprocity. This may necessitate parallel governance structures: deeper cooperation within democratic alliances alongside more limited technical coordination with strategic competitors on narrow, clearly defined challenges like AI safety research and incident notification.

SITA's 75-year evolution from 11 airlines sharing telecommunications costs to a $1.5 billion global cooperative serving 90% of world aviation demonstrates that industry-owned infrastructure for safety-critical technology can achieve remarkable scale and longevity. Whether AI governance can replicate this success depends not only on institutional design but on broader geopolitical conditions shaping possibilities for cooperation. SITA succeeded by combining economic necessity, safety imperatives, technical focus, and cooperative ownership within a favorable political context. AI governance advocates must attend to all these elements—and recognize when political realities impose constraints that technical ingenuity alone cannot overcome.

‍

Bibliography: Bridging Divides - A Practical Framework for Global AI Governance

International Organizations & Governance Bodies

Brookings Institution

• "Strengthening International Cooperation on AI" - https://www.brookings.edu/articles/strengthening-international-cooperation-on-ai/
• "The Bletchley Park Process Could Be a Building Block for Global Cooperation on AI Safety" - https://www.brookings.edu/articles/the-bletchley-park-process-could-be-a-building-block-for-global-cooperation-on-ai-safety/
• "Network Architecture for Global AI Policy" - https://www.brookings.edu/articles/network-architecture-for-global-ai-policy/
• "Steps Toward AI Governance in the Military Domain" - https://www.brookings.edu/articles/steps-toward-ai-governance-in-the-military-domain/

Carnegie Endowment for International Peace

• "DeepSeek and Other Chinese Firms Converge with Western Companies on AI Promises" (January 2025) - https://carnegieendowment.org/research/2025/01/deepseek-and-other-chinese-firms-converge-with-western-companies-on-ai-promises
• "How the UK Should Engage China at AI's Frontier" (October 2024) - https://carnegieendowment.org/posts/2024/10/lammy-china-ai-safety-cooperation
• "A CERN Model for Studying the Information Environment" (November 2022) - https://carnegieendowment.org/research/2022/11/a-cern-model-for-studying-the-information-environment
• "India's AI Strategy: Balancing Risk and Opportunity" (February 2024) - https://carnegieendowment.org/posts/2024/02/indias-ai-strategy-balancing-risk-and-opportunity

Center for Strategic and Emerging Technology (CSET), Georgetown University

• "From Cold War Sanctions to Weaponized Interdependence" - https://cset.georgetown.edu/publication/from-cold-war-sanctions-to-weaponized-interdependence/
• "Securing Global Standards for Innovation and Growth" - https://www.csis.org/analysis/securing-global-standards-innovation-and-growth

OECD (Organisation for Economic Co-operation and Development)

• "Regulatory Sandboxes Can Facilitate Experimentation in Artificial Intelligence" - https://oecd.ai/en/wonk/sandboxes
• "Singapore's Model Framework Balances Innovation and Trust in AI" - https://oecd.ai/en/wonk/singapores-model-framework-to-balance-innovation-and-trust-in-ai
• "Technology Governance" - https://www.oecd.org/en/topics/sub-issues/technology-governance.html

UNESCO (United Nations Educational, Scientific and Cultural Organization)

• "Ethics of Artificial Intelligence" - https://en.unesco.org/artificial-intelligence/ethics
• "UNESCO's Recommendation on the Ethics of Artificial Intelligence: Key Facts" - https://www.unesco.org/en/articles/unescos-recommendation-ethics-artificial-intelligence-key-facts
• "UNESCO Recommendation on the Ethics of Artificial Intelligence" - https://dig.watch/resource/unesco-recommendation-on-the-ethics-of-artificial-intelligence
• "UNESCO Launches Global AI Ethics and Governance Observatory" (2024) - https://digital-skills-jobs.europa.eu/en/latest/news/unesco-launches-global-ai-ethics-and-governance-observatory-2024-global-forum-ethics

United Nations

• "Secretary-General's Remarks at High-level Multi-Stakeholder Informal Meeting to Launch the Global Dialogue on Artificial Intelligence Governance" (September 2025) - https://www.un.org/sg/en/content/sg/statement/2025-09-25/secretary-generals-remarks-high-level-multi-stakeholder-informal-meeting-launch-the-global-dialogue-artificial-intelligence-governance-delivered
• "Global Dialogue on Artificial Intelligence Governance" - https://press.un.org/en/2025/sgsm22839.doc.htm
• "COP29: Digital Tech and AI Can Boost Climate Action" (November 2024) - https://news.un.org/en/story/2024/11/1157086

RAND Corporation

• "Winning the Tech Cold War" (August 2023) - https://www.rand.org/pubs/commentary/2023/08/winning-the-tech-cold-war.html
• "Full Stack: China's Evolving Industrial Policy for AI" - https://www.rand.org/pubs/perspectives/PEA4012-1.html

Center for Security and Emerging Technology (CSIS)

• "What the UN Global Dialogue on AI Governance Reveals About Global Power Shifts" - https://www.csis.org/analysis/what-un-global-dialogue-ai-governance-reveals-about-global-power-shifts

Academic Research & Publications

Nature (Scientific Journal)

• "China and the U.S. Produce More Impactful AI Research When Collaborating Together" (November 2024) - https://www.nature.com/articles/s41598-024-79863-5
• "What CERN Does Next Matters for Science and for International Cooperation" - https://www.nature.com/articles/d41586-025-00787-9

University of Chicago Knowledge Archive

• "China and the U.S. Produce More Impactful AI Research When Collaborating Together" - https://knowledge.uchicago.edu/record/14076

Oxford Academic (International Affairs)

• "Global AI Governance: Barriers and Pathways Forward" - https://academic.oup.com/ia/article/100/3/1275/7641064
• "The Geopolitics of Technology Standards: Historical Context for US, EU and Chinese Approaches" - https://academic.oup.com/ia/article/100/4/1635/7692873

Springer

• "Charting the Technology Cold War: The Balance of Loss, U.S. Containment Strategies and China's Countermeasures" - https://link.springer.com/article/10.1007/s42533-025-00180-1

ScienceDirect

• "The Influence of China in AI Governance Through Standardisation" - https://www.sciencedirect.com/science/article/abs/pii/S0308596123001842
• "Privacy Preservation in Federated Learning: An Insightful Survey from the GDPR Perspective" - https://www.sciencedirect.com/science/article/pii/S0167404821002261

ResearchGate

• "Model Cards for Model Reporting" - https://www.researchgate.net/publication/330268857_Model_Cards_for_Model_Reporting
• "Datasheets for Datasets" - https://www.researchgate.net/publication/324055506_Datasheets_for_Datasets

arXiv

• "Audit Cards: Contextualizing AI Evaluations" - https://arxiv.org/html/2504.13839v1
• "Publication Trends in Artificial Intelligence Conferences: The Rise of Super Prolific Authors" - https://arxiv.org/html/2412.07793v1

Preprints.org

• "Advancing Privacy-Preserving AI: A Survey on Federated Learning and Its Applications" (January 2025) - https://www.preprints.org/manuscript/202501.0685

SSRN (Social Science Research Network)

• "Artificial Intelligence and Global Security: Strengthening International Cooperation and Diplomatic Relations" - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4998408

Medium / Substack Publications

• "Running Code and Rough Consensus for AI: Polycentric Governance in the Algorithmic Age" - https://medium.com/@AdamThierer/running-code-and-rough-consensus-for-ai-polycentric-governance-in-the-algorithmic-age-f6691471a804
• "Forked by Regulation: The Reality of Building AI for China vs. America" - https://medium.com/@collin.a.spears/forked-by-regulation-the-reality-of-building-ai-for-china-vs-america-4728c61f3559
• "Federated Learning: Decentralized Machine Learning for Privacy-Preserving AI" - https://medium.com/@rahulholla1/federated-learning-decentralized-machine-learning-for-privacy-preserving-ai-3601282c8462
• "AI & Health in China: An Essential Overview" - https://chinahealthpulse.substack.com/p/ai-and-health-in-china-an-essential

Regional AI Governance Frameworks

Singapore

• Singapore Personal Data Protection Commission: "Singapore's Approach to AI Governance" - https://www.pdpc.gov.sg/help-and-resources/2020/01/model-ai-governance-framework
• IMDA: "Model AI Governance Framework 2024" - https://www.imda.gov.sg/resources/press-releases-factsheets-and-speeches/press-releases/2024/public-consult-model-ai-governance-framework-genai
• White & Case: "AI Watch: Global Regulatory Tracker - Singapore" - https://www.whitecase.com/insight-our-thinking/ai-watch-global-regulatory-tracker-singapore

India

• Observer Research Foundation: "AI Governance in India: Aspirations and Apprehensions" - https://www.orfonline.org/research/ai-governance-in-india-aspirations-and-apprehensions
• Digital Watch Observatory: "India's National Strategy for Artificial Intelligence" - https://dig.watch/resource/indias-national-strategy-for-artificial-intelligence
• National Bureau of Asian Research: "AI Governance in India" - https://www.nbr.org/publication/ai-governance-in-india/

United Arab Emirates

• UAE Legislation: "UAE Charter for the Development and Use of AI" - https://uaelegislation.gov.ae/en/policy/details/the-uae-charter-for-the-development-and-use-of-artificial-intelligence
• Bird & Bird: "GCC Navigating AI Regulations - The Current Landscape" - https://www.twobirds.com/en/insights/2025/united-arab-emirates/gcc-navigating-ai-regulations---the-current-landscape
• Modulos: "Navigating AI Governance in the UAE" - https://www.modulos.ai/blog/uae-ai-regulations-guide/

European Union

• EU Artificial Intelligence Act: "AI Regulatory Sandbox Approaches: EU Member State Overview" - https://artificialintelligenceact.eu/ai-regulatory-sandbox-approaches-eu-member-state-overview/
• EU Artificial Intelligence Act: "Article 57: AI Regulatory Sandboxes" - https://artificialintelligenceact.eu/article/57/
• DigiCon: "Regulatory Sandboxes in the AI Act Between Innovation and Safety" - https://digi-con.org/regulatory-sandboxes-in-the-ai-act-between-innovation-and-safety/
• European Commission: "Mutual Recognition Agreements" - https://single-market-economy.ec.europa.eu/single-market/goods/international-aspects/mutual-recognition-agreements_en
• European Medicines Agency: "Mutual Recognition Agreements (MRA)" - https://www.ema.europa.eu/en/human-regulatory-overview/research-development/compliance-research-development/good-manufacturing-practice/mutual-recognition-agreements-mra

China

• Reed Smith: "Navigating the Complexities of AI Regulation in China" (August 2024) - https://www.reedsmith.com/en/perspectives/2024/08/navigating-the-complexities-of-ai-regulation-in-china
• China Briefing: "China's DeepSeek and its Open-Source AI Models" - https://www.china-briefing.com/news/chinas-deepseek-and-its-open-source-ai-models/

United States

• The Regulatory Review: "The United States Regulates Artificial Intelligence with Export Controls" - https://www.theregreview.org/2025/09/25/flatley-the-united-states-regulates-artificial-intelligence-with-export-controls/
• Export Compliance Daily: "Without Changes, Biden's AI Export Control Rule Will Be 'Gift' to China, Microsoft Says" - https://exportcompliancedaily.com/article/2025/02/28/without-changes-bidens-ai-export-control-rule-will-be-gift-to-china-microsoft-says-2502270074
• NIST: "The UK-US Blog Series on Privacy-Preserving Federated Learning: Introduction" - https://www.nist.gov/blogs/cybersecurity-insights/uk-us-blog-series-privacy-preserving-federated-learning-introduction
• NIST: "Mutual Recognition Agreements for Conformity Assessment of Telecommunications Equipment" - https://www.nist.gov/standardsgov/mutual-recognition-agreements-conformity-assessment-telecommunications-equipment
• FCC: "Equipment Authorization - Mutual Recognition Agreements" - https://www.fcc.gov/general/equipment-authorization-mutual-recognition-agreements

Technical Standards & Industry Organizations

ISO/IEC Standards

• JTC 1 Standards Information: "SC 42 - Artificial Intelligence" - https://jtc1info.org/sd-2-history/jtc1-subcommittees/sc-42/
• ISO: "ISO/IEC JTC 1/SC 42 - Artificial intelligence" - https://www.iso.org/committee/6794475.html

IEEE (Institute of Electrical and Electronics Engineers)

• IEEE Standards on AI ethics and technical frameworks

ITU (International Telecommunication Union)

• Digital Commons: "The International Telecommunications Union: 130 Years of..." - https://digitalcommons.du.edu/context/djilp/article/1717/viewcontent/24_23DenvJIntlL_Poly501_1994_1995_.pdf

Open Source & Chinese AI Development

Interconnects.ai

• "Ranking the Chinese Open Model Builders" - https://www.interconnects.ai/p/chinas-top-19-open-model-labs
• "On China's Open Source AI Trajectory" - https://www.interconnects.ai/p/on-chinas-open-source-ai-trajectory

TechCrunch

• "Hugging Face CEO Has Concerns About Chinese Open Source AI Models" (December 2024) - https://techcrunch.com/2024/12/03/huggingface-ceo-has-concerns-about-chinese-open-source-ai-models/

The Diplomat

• "Is China Building the Future of AI Governance Through Open-Source Modeling?" - https://thediplomat.com/2025/05/is-china-building-the-future-of-ai-governance-through-open-source-modeling/
• "The Rise of AI Manufacturing in China and South Korea" - https://thediplomat.com/2025/05/the-rise-of-ai-manufacturing-in-china-and-south-korea/

The Conversation

• "DeepSeek: How China's Embrace of Open-Source AI Caused a Geopolitical Earthquake" - https://theconversation.com/deepseek-how-chinas-embrace-of-open-source-ai-caused-a-geopolitical-earthquake-249563

South China Morning Post

• "How China's Open-Source AI Is Helping DeepSeek, Alibaba Take on Silicon Valley" - https://www.scmp.com/tech/big-tech/article/3318747/how-chinas-open-source-ai-helping-deepseek-alibaba-take-silicon-valley

Rest of World

• "US and China Lead Global AI Collaboration as Research Partnerships Grow" - https://restofworld.org/2025/us-china-lead-global-ai-collaboration/

ITIF (Information Technology and Innovation Foundation)

• "How Innovative Is China in AI?" (August 2024) - https://itif.org/publications/2024/08/26/how-innovative-is-china-in-ai/

Privacy-Preserving Technologies & Federated Learning

Roundtable Data Science Salon

• "Federated Learning for Privacy-Preserving AI: An In-Depth Exploration" - https://roundtable.datascience.salon/federated-learning-for-privacy-preserving-ai-an-in-depth-exploration

Netguru

• "Federated Learning: A Privacy-Preserving Approach to Collaborative AI Model Training" - https://www.netguru.com/blog/federated-learning

Palo Alto Networks

• "What Is Federated Learning? A Guide to Privacy-Preserving AI" - https://www.paloaltonetworks.com/cyberpedia/what-is-federated-learning

STL Partners

• "Federated Learning: Decentralised Training for Privacy-Preserving AI" - https://stlpartners.com/articles/edge-computing/federated-learning/

Science Diplomacy & Historical Cooperation Models

Science & Diplomacy Journal

• "Past, Present, and Future of Science Diplomacy in Europe" (2018) - https://www.sciencediplomacy.org/perspective/2018/past-present-and-future-science-diplomacy-in-europe
• "CERN and SESAME – Science Diplomacy Building Bridges" (2022) - https://www.sciencediplomacy.org/perspective/2022/cern-and-sesame-science-diplomacy-building-bridges
• "Learning from Success: Lessons in Science and Diplomacy from the Montreal Protocol" (2020) - https://www.sciencediplomacy.org/article/2020/learning-success-lessons-in-science-and-diplomacy-montreal-protocol

Shells and Pebbles

• "CERN: The Foundational Myth of European Science Diplomacy" - https://www.shellsandpebbles.com/2025/03/04/cern-the-foundational-myth-of-european-science-diplomacy/

CERN

• "CERN Speaks at UN About the Laboratory's Cooperation Model" - https://www.cern/news/news/cern/cern-speaks-un-about-laboratorys-cooperation-model
• "Partnerships for the Goals - International Relations" - https://international-relations.web.cern.ch/stakeholder-relations/international-organizations/partnerships-goals

LinkedIn

• "How CERN Is Advancing Science Diplomacy" - https://www.linkedin.com/pulse/how-cern-advancing-science-diplomacy-yulia-fedorenko

Wikipedia

• "Montreal Protocol" - https://en.wikipedia.org/wiki/Montreal_Protocol
• "Mutual Recognition" - https://en.wikipedia.org/wiki/Mutual_recognition
• "X.25" - https://en.wikipedia.org/wiki/X.25
• "Public Data Network" - https://en.wikipedia.org/wiki/Public_data_network
• "Aeronautical Telecommunication Network" - https://en.wikipedia.org/wiki/Aeronautical_Telecommunication_Network

Regulatory Sandboxes & Innovation Frameworks

Future of Privacy Forum

• "Balancing Innovation and Oversight: Regulatory Sandboxes as a Tool for AI Governance" - https://fpf.org/blog/balancing-innovation-and-oversight-regulatory-sandboxes-as-a-tool-for-ai-governance/

European Parliament Think Tank

• "Artificial Intelligence [What Think Tanks Are Thinking]" (March 2024) - https://epthinktank.eu/2024/03/27/artificial-intelligence-what-think-tanks-are-thinking-5/

GDPR Local

• "International AI Regulations - Current AI Landscape" - https://gdprlocal.com/international-ai-regulations-current-ai-landscape/

Healthcare AI & Sector-Specific Applications

CKGSB (Cheung Kong Graduate School of Business)

• "AI Applications in China's Healthcare System" - https://english.ckgsb.edu.cn/knowledge/article/ai-applications-in-china-healthcare-system/

Eularis

• "FDA Approved AI: Where Are We Today?" - https://eularis.com/fda-approved-ai-where-are-we-today/

Times Higher Education

• "Can China's Universities Power It to Victory in the Global AI Race?" - https://www.timeshighereducation.com/depth/can-chinas-universities-power-it-victory-global-ai-race

Aviation & SITA-Specific References

SITA (Official Sources)

• "SITA Celebrates 75 Years of Innovation and Leadership in Air Transport Technology" - https://www.sita.aero/pressroom/news-releases/sita-celebrates-75-years-of-innovation-and-leadership--in-air-transport-technology/
• "SITA Board" - https://www.sita.aero/about-us/meet-the-team/sita-board/
• "SITA Council" - http://www.sita.aero/about-us/sita-board-council-members/sita-council
• "Chair's Statement 2022" - https://www.sita.aero/sita-activity-report-2022/executive-statements/chair-statement/
• "Council Representatives 2020" - https://www.sita.aero/sita-activity-report-2020/council-representatives/
• "SITA Reports Record Results" - https://www.sita.aero/pressroom/news-releases/sita-reports-record-results-as-it-pursues--a-strategy-of-fast-growth-and-expansion/
• "Aeroflot Deploys SITA CrewTablet" - https://www.sita.aero/pressroom/news-releases/aeroflot-deploys-sita-crewtablet
• "SITA Group Consolidated Financial Statements 2023" (PDF) - https://www.sita.aero/contentassets/39bbb0cbadf14dd4951e3c81ec2c0cc2/sita-group-consolidated-financial-statements-2023.pdf

ICAO (International Civil Aviation Organization)

• "SITA - The Postal History of ICAO" - https://applications.icao.int/postalhistory/sita_societe_internationale_de_telecommunications_aeronautiques.htm
• SKYbrary Aviation Safety: "Chicago Convention" - https://skybrary.aero/articles/chicago-convention

IATA (International Air Transport Association)

• "Type B Messaging Whitepaper Version 2.1" (June 2024) - https://www.iata.org/contentassets/badbfd2d36a74f12b021c9dd899ecbad/type_b_messaging_whitepaper_v2dot1_14_june_2024.pdf
• "Type B Messaging Whitepaper V2.5" (May 2025) - https://www.iata.org/contentassets/badbfd2d36a74f12b021c9dd899ecbad/type-b-messaging-whitepaper-v2.5.pdf

Academic & Historical Sources on Aviation

• SMU Scholar: "Aeroflot, The Soviet Airline - At Home and Abroad" - https://scholar.smu.edu/cgi/viewcontent.cgi?article=2443&context=jalc
• Michigan State University: "Aeroflot and Pan Am – Seventeen Moments in Soviet History" - https://soviethistory.msu.edu/1968-2/aeroflot-and-pan-am/
• Key Aero: "Eastern Bloc Party: Part One" - https://www.key.aero/article/eastern-bloc-party-part-one
• Henry Tenby: "Iron Curtain Airlines of the 1970s - A Photo History" - https://www.henrytenby.com/iron-curtain-airlines-of-the-1970s-a-photo-history/

Wikipedia

• "SITA (Business Services Company)" - https://en.wikipedia.org/wiki/SITA_(business_services_company)
• "International Telecommunication Union" - https://en.wikipedia.org/wiki/International_Telecommunication_Union

Industry Publications

• FundingUniverse: "History of Equant N.V." - https://www.fundinguniverse.com/company-histories/equant-n-v-history/
• Airport Technology: "SITA" - https://www.airport-technology.com/contractors/baggage/spirit-media-ltd-obo-sita/
• Airports International: "SITA Celebrates First 75 Years" - https://www.airportsinternational.com/article/sita-celebrates-first-75-years
• Computer Weekly: "SITA Takes Off with Cathay Pacific to Expand Global Network Connectivity" - https://www.computerweekly.com/news/366618499/SITA-takes-off-with-Cathay-Pacific-to-expand-global-network-connectivity
• Orange Business: "SITA: Business Solutions for the Air Transport Industry" - https://www.orange-business.com/en/partners/sita
• CanvasBusinessModel: "What is Brief History of SITA Company?" - https://canvasbusinessmodel.com/blogs/brief-history/sita-brief-history
• CanvasBusinessModel: "Who Owns SITA" - https://canvasbusinessmodel.com/blogs/owners/sita-who-owns
• CanvasBusinessModel: "What is Competitive Landscape of SITA Company?" - https://canvasbusinessmodel.com/blogs/competitors/sita-competitive-landscape

Aviation Safety & AI Governance Comparisons

UC Berkeley Center for Long-Term Cybersecurity

• "The Flight to Safety-Critical AI: Lessons in AI Safety from the Aviation Industry" (2024) - https://cltc.berkeley.edu/publication/new-report-the-flight-to-safety-critical-ai-lessons-in-ai-safety-from-the-aviation-industry/

Medium

• "Aviation to AI: How to Regulate Artificial Intelligence?" - https://medium.com/sdg-counting/aviation-to-ai-how-to-regulate-artificial-intelligence-7acad56b282b

Aero

• "International Aviation Regulations - A Full Guide" - https://an.aero/international-aviation-regulations/

Additional Technical & Trade Resources

ICANN

• "Summary of Application of Societe Internationale de Telecommunications Aeronautiques" - https://archive.icann.org/en/tlds/report/air1.html

Lark

• "Mutual Recognition Agreement (MRA)" - https://www.larksuite.com/en_us/topics/quality-management-glossary/mutual-recognition-agreement-mra

Stack Exchange

• "What Were the Major Things That Caused TCP/IP to Become the Internet Standard Protocol?" - https://retrocomputing.stackexchange.com/questions/29823/what-were-the-major-things-that-caused-tcp-ip-to-become-the-internet-standard-pr

Quartz

• "The Data That Changed the Direction of AI Research—and Possibly the World" - https://qz.com/1034972/the-data-that-changed-the-direction-of-ai-research-and-possibly-the-world

Taylor & Francis Online

• "Governing the Digital Economy: Transatlantic Accommodation and Cooperation" - https://www.tandfonline.com/doi/full/10.1080/07036337.2024.2398429

‍

Note: URLs were current as of December 2025. Some sources may require institutional access or subscriptions.

‍

‍