Skyline Ventures

Unsurprisingly, AI Goverance has its fair share of security risks. Right now, there's no "zero-trust equivalent" for cross-platform AI. This makes workflows easier to penetrate and data easier to access, potentially leading to cascading failures across the entire system. In an interoperable environment, agents constantly make requests across systems, vendors, and data sources. Every one of those requests potentially exposes a vulnerability. A zero-trust approach doesn't assume the safety of any AI agent or system, meaning it verifies every interaction against governance and security policies before moving to the next step. Enforce least-privilege access for every agent, build modular and containerized, and narrowly score credentials.

• Interoperability of governance approaches
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Safe, secure and trustworthy AI
• Open-source software, open data and open AI models

AI interoperability can also create serious accountability and governance gaps. It's hard to know who "owns" an error when multiple systems from multiple vendors are involved. Audit trails are more difficult to clearly decipher, data sovereignty becomes muddled, and iron-clad permissioning becomes more essential than ever. Deploying an overarching orchestration layer that coordinates tasks, manages dependencies, and handles agent-to-agent communication. Pair this layer with advanced telemetry and comprehensive observability. Widely considered to be the most serious roadblock to AI interoperability (especially at scale), protocol fragmentation between MCP, A2A, ACP, and other frameworks means that no clear "standard of communication" between AI agents has emerged. In my view, it essentially re-creates the "frankenstack" problem of disconnected third-party integrations.

AI interoperability is the ability for different AI Agents, models, and systems to autonomously collaborate across sectors and take collective action to achieve desired outcomes, even if those AI agents are from different vendors, platforms, or domains. Orchestration is the conductor that coordinates how agents, systems, and tools work together to achieve their shared goal. Orchestration breaks down larger objectives into specific sub-tasks, assigns specific actions and tasks to specific tools or agents, and manages the data handoffs between them. AI Orchestration executes complex, multi-step, and cross-functional workflows end-to-end - all while ensuring compliance and governance.

AI interoperability alongside the governance is essential for today's enterprise workforce. It eliminates data and application silos, prevents single vendor lock-in, autonomously executes complicated workflows, and provides a seamless customer experience. 75% of organizations plan to deploy multi-agent frameworks within the next 18 months, and 45% of organizations that have already scaled AI agents are actively piloting multi-agent systems instead of single-use bots in Europe. Multi-agent AI systems successfully complete 70% more processes than individual AI agents, deliver 40% faster execution, and have 25% lower operating costs compared to traditional workflow processes. 60% of multi-agent systems are expected to deploy and supported by AI governance by 2028.

As AI Dialogue advances while applying international cooperation on AI governance, it increasingly recognise the need for international governance to address shared benefits and challenges. However, international cooperation is complex and costly, and not all AI issues require cooperation at the international level. A framework is required to identify and prioritise AI governance issues warranting internationalisation within any organisation. We analyse critical policy areas across data, compute, and model governance using four factors which broadly incentivise states to internationalise governance efforts: cross-border externalities, regulatory arbitrage, uneven governance capacity, and interoperability. We find strong benefits of internationalisation in compute-provider oversight, content provenance, model evaluations, incident monitoring, and risk management protocols. In contrast, the benefits of internationalisation are lower or mixed in data privacy, data provenance, chip distribution, and bias mitigation. These results can guide policymakers and researchers in prioritising international AI governance efforts.

In 2023, both the US and China signed the Bletchley Declaration, acknowledging the potential for serious harm from advanced AI systems as well as the importance of cooperating to mitigate it. Since then, the US and China have also jointly signed UN resolutions on AI issues, held a round of intergovernmental dialogues on AI, and agreed to limit the integration of AI into control systems for nuclear weapons. As AI dialogue and governance becomes increasingly central to global policymaking, ensuring robust digital rights frameworks is essential to fostering public trust, equitable access, and regulatory effectiveness. In recent years, digital rights have been formalized through national and regional instruments, including the European Union's Declaration of Digital Rights and Principles and South Korea's Digital Bill of Rights. These frameworks provide a comprehensive foundation to guide legislative, policy, and technological development. In 2024, as the United Nations commits to a Global Digital Compact, there is renewed hope for more forward-thinking, globally coordinated policymaking that safeguards citizens' rights in the digital age. With the recent formation of two AI industry consortia—the Frontier Model Forum (FMF) and the US AI Safety Institute Consortium (AISIC)—it is increasingly important to understand how voluntary safety initiatives can mitigate AI risks with their AI dialogues without imposing excessive burdens on firms.

Different stakeholders can use Adaptive and Interactive Event Formats, where AI is being used to transform traditional events into continuous, two-way dialogues through generative tools. Via Reaction Live: Monitors audience emotions and feedback in real-time, moving beyond static surveys. On AI Quiz & Polls: Intelligent trivia generated instantly from live speech or panel content to maintain audience attention. Interactive Walls: Digital mosaics that transform participant messages or thoughts into evolving collective art. Finally, AI Co-hosts: Digital avatars that moderate panels, read audience messages, and facilitate complex sessions. In these measures, they can contribute to the AI Dialogue. The other format and structure of thee AI Dialogue are Multimodal and Accessibility-Driven Formats. They are Adaptive Content Summaries on Generating short video summaries or visual maps from complex technical dialogues to suit different learning styles. Real-Time Translation is Breaking cultural barriers by providing instant dialogue translation for multicultural participants. And finally, Emotionally Intelligent Feedback on the Systems that detect vocal tone or facial expressions in virtual meetings to adjust the AI's response or alert human leaders to team morale.

Global discussions on AI governance are heavily dominated by high-income nations and large technology firms, leading to a significant 'perspective deficit'. This creates a, top-down governance model that marginalizes the needs and realities of a large portion of the world's population, often referred to as the 'Global AI Majority', which then imposes concerns on global discussions on AI governance. Marginalized and Indigenous Communities includes Indigenous Populations: The lack of digital tools for language preservation and the underutilization of traditional knowledge in AI systems make these communities acutely underrepresented. Rural Regions & Informal Economies: Public services and economic opportunities are optimized for urban, digital-first populations, overlooking the needs of those in rural or informal economies. Vulnerable & Conflict-Affected Populations: People in conflict zones or unstable situations are often disproportionately harmed by surveillance and automated decision-making but have no voice in setting guidelines. Sociotechnical and Intersectional underrepresented Perspectives Women, Particularly in Tech Development: Women make up only roughly 30% of the AI workforce globally. This lack of representation in development teams leads to biased systems that disadvantage women in recruitment, credit scoring, and public services. Minority Practitioners: The perspectives of practitioners working within marginalized communities in high-income countries are often overlooked. Age-Diverse Voices: Both the very young and the elderly are systematically underrepresented in policy forums, despite being heavily impacted by educational and care-related AI. There is a lack of focus on community-led, cooperative, or open-source governance structures, in favor of corporate or governmental regulation.

Innovative engagement formats in AI dialogues are shifting from passive, one-off Q&A to interactive, collaborative, and persistent experiences that feel human and act in real-time. Key trends include using AI as a co-creator, adapting conversations to user sentiment, and integrating AI within existing workflows. The other innovative enagagement formats within AI governance are Immersive Educational and Training Simulations, where New formats act major play in learning focus on active problem-solving and personalized role-play. This combines with Reaction Live: Monitors audience emotions and feedback in real-time, moving beyond static surveys. AI Quiz & Polls: Intelligent trivia generated instantly from live speech or panel content to maintain audience attention. Interactive Walls: Digital mosaics that transform participant messages or thoughts into evolving collective art. AI Co-hosts: Digital avatars that moderate panels, read audience messages, and facilitate complex sessions. The least, but not least Adaptive and Interactive Event Formats are where AI is being used to transform traditional events into continuous, two-way dialogues through generative tools. Scenario Simulations: AI-driven "choose-your-own-adventure" style dialogues for exploring the impacts of community or business decisions. AI Challenger Sessions: Inviting an AI into problem-solving meetings specifically to act as a "devil's advocate" or challenger. Role-Play Training: Using AI to generate realistic, multi-character scripts—such as a project manager negotiating with a demanding client—which can then be performed or analyzed by learners. Case Study Gamification: Transforming text-based case studies into interactive word walls, Jeopardy boards, or scavenger hunts.

NIST AI Risk Management Framework (AI RMF): A voluntary, flexible framework focused on mapping, measuring, managing, and governing AI risks. ISO/IEC 42001:2023: A significant international standard offering a structured framework for AI management systems. OECD AI Principles: Principles promoting trustworthy AI that respects human rights, updated in 2024 for practical application. Internal AI Policies: Organizations should adopt clear AI policies defining acceptable use, prohibited use cases, and responsible parties. EU AI Act Compliance: Adopting a risk-based approach (low to high risk) that mandates specific compliance requirements.