Key Facts
The ADGM Registration Authority’s Emerging Technologies function and Hashed have jointly published “Web3 Leaders Roundtable”, a policy report from a closed-door session held during Abu Dhabi Finance Week 2025.
The report identifies three core findings: financial infrastructure is moving toward machine-native execution, governance and evidentiary integrity are becoming binding requirements for AI, and tokenization has shifted from an issuance question to a market-structure question.
Participants flagged Basel risk-weighted asset charges of up to 1,250% on certain digital-asset exposures, fragmented FATF travel rule implementation, and unresolved accounting and tax treatment as compounding constraints on institutional deployment.
DTCC’s Nadine Chakar told the roundtable the firm aspires to tokenize the entirety of U.S. capital markets — a stated objective rather than an active program.
Closed-door participants included Joseph Lubin (Consensys), Joseph Chalom (Sharplink), Tony Ashraf (BlackRock), Michael Reed (Franklin Templeton), Alexander Lipton (ADIA), Lily Liu (Solana Foundation), Evy Theunis (DBS), Matt Long (FalconX), Marco Dal Lago (Tether), Kash Razzaghi (Circle), Peter Kerstens (European Commission), and Clara Guerra (Government of Liechtenstein).
Tokenization has moved from an issuance question to a market-structure question, with settlement design, collateral mobility, identity, and prudential treatment now setting the ceiling on institutional adoption. That is the central finding of a joint policy report published by the ADGM Registration Authority’s Emerging Technologies function and Hashed, drawing on a closed-door Web3 Leaders Roundtable held during Abu Dhabi Finance Week 2025.
The roundtable was convened by Dmitry Fedotov, Head of Emerging Technologies at the ADGM Registration Authority, and Simon Kim, Chief Executive Officer and Managing Partner of Hashed. Welcoming remarks were delivered by H.E. Rashed Al Blooshi, Chief Executive Officer of the ADGM Registration Authority, and H.E. Ahmed Bin Sulayem, Executive Chairman and Chief Executive Officer of the Dubai Multi Commodities Centre.
Three core findings on AI and on-chain finance
The report’s executive brief sets out three takeaways. First, financial infrastructure is likely to evolve toward machine-native execution and settlement, as autonomous AI agents drive transaction volumes that legacy internet and payment systems were not designed to support. Second, governance and evidentiary integrity — tamper-evident records of data lineage, model versions and signed agent actions — are likely to become binding requirements rather than optional features for AI systems acting in financial and physical environments.
Third, tokenization scalability now depends on alignment across legal rights, market infrastructure, and supervisory implementation, not the mechanics of issuance. The report describes a “compounding” constraint set across prudential capital treatment, custody and recoverability expectations, fragmented anti-money laundering and know-your-customer implementation, and unresolved accounting and tax treatment.
Session 1: Blockchain as the trust layer for AI agents
The first session, moderated by Henri Arslanian of Nine Blocks Capital, framed blockchain as the coordination substrate for AI-mediated economies. Joseph Lubin of Consensys argued that the settlement infrastructure for machine-native interaction already exists, citing Ethereum’s capacity for fractional-cent and cent-level settlement at volumes infeasible on traditional rails.
Jan Liphardt of OpenMind, drawing on the firm’s deployment of humanoid robots in public environments, described immutability as foundational for verifying what occurred and under what authority. Chi Zhang of Kite AI argued that AI agents accessing data sources are routinely treated as malicious bots by current web infrastructure, forcing reliance on rate limits, API keys, or wholesale licensing — friction that blockchain-based payment and access mechanisms can resolve.
The discussion of governance focused on what the report calls the “digital twin” — a personal AI agent operating with persistent memory and defined objectives on behalf of an individual. Lubin and Illia Polosukhin of Near Protocol framed universal access to such agents as a structural-equity question rather than a premium feature, warning against re-creating the centralised extraction models of earlier internet eras. Rayhaneh Sharif-Askary of Grayscale described decentralisation as a prerequisite for fair and transparent AI systems.
Alexander Lipton of the Abu Dhabi Investment Authority cautioned that comprehensive alignment is currently unachievable. He proposed beginning with narrowly scoped agents tied to measurable objectives, allowing oversight to evolve incrementally.
Identity, reputation, and risk-segmented governance
Participants converged on the limitations of static, document-based identity in agent-driven environments. Joseph Chalom of Sharplink argued that markets will revert toward reputation-based models augmented by AI’s capacity to assess risk dynamically — a shift from “good wallet versus bad wallet” classification to systems where agents themselves build, degrade, and transfer reputation over time.
Xin Song of GSR set out a risk-segmented governance approach in which obligations scale with the impact of an agent’s activity. Liability frameworks, Song argued, are converging toward role-based accountability, distributing responsibility across developers (defective models), deployers (unsafe deployment), and operators (negligent oversight).
Kash Razzaghi of Circle warned that AI-driven systems will increase transaction volumes by orders of magnitude that regulation alone cannot absorb, arguing that trust must precede adoption and adoption must precede effective regulation.
Session 2: What is blocking institutional scale
The second session, moderated by Hashed General Partner Baek Kim with framing remarks from Hashed Global Head of Legal Jin Kang, concentrated on operational blockers to institutional adoption. The session catalogued four parallel product structures: tokenisation as a distribution mechanism for traditional instruments, stablecoins as settlement infrastructure, liquid staking as a liquidity tool, and digital asset treasuries as balance-sheet strategies.
Michael Reed of Franklin Templeton outlined a three-pillar approach combining ledger-efficiency investment, crypto as an investable asset class, and direct validator participation. Thomas Uhm of the Jito Foundation reframed liquid staking tokens as a redemption-uncertainty solution rather than a yield product — solving the problem of protocol-level unstaking queues that complicate ETF redemption mechanics. Lily Liu of the Solana Foundation positioned digital asset treasuries as a path toward operating businesses built around networks, distinguishing durable models from short-lived token wrappers.
Where regulation is binding
Participants identified several specific friction points. Jez Mohideen of Laser Digital described on-chain transactions passing through two to three overlapping layers of know-your-customer checks across custodians, counterparties, and regulated entities, even where Financial Action Task Force standards are already met. Chris Rayner-Cook of Brevan Howard Digital framed the core question as whether accounting and capital frameworks will permit stablecoins to function as neutral settlement assets on institutional balance sheets.
Matt Long of FalconX flagged inconsistent custody expectations across jurisdictions and Basel risk-weighted asset treatment as binding constraints on market intermediation, limiting firms’ ability to warehouse inventory and extend liquidity. Chalom argued the primary obstacle to scaling tokenisation is the absence of a standardised, regulator-supported framework for institutional digital identity, with secondary trades still restricted to whitelisted counterparties under issuer-specific onboarding.
Peter Kerstens, an Advisor at the European Commission, situated the EU’s parallel MiCA framework in the practical context of reconciling 27 member states within an existing dense body of securities law. Clara Guerra of the Government of Liechtenstein argued for treating tokens as containers for legal rights and regulating economic function rather than business models. Both warned that supervisory posture and implementation consistency matter as much as legal drafting.
Tokenised equities and disclosure
The session also surfaced a market-conduct concern around tokenised equities. Participants warned that retail users encountering products such as “tokenised Tesla” may assume shareholder rights, when in practice the instrument can be a derivative wrapper with weekend liquidity gaps and no voting rights. Standardised identifiers, labelling conventions, and legal-rights classification were framed as foundational market-structure issues, not consumer-education additions.
Report: https://assets.adgm.com/download/assets/ADFW+2025+Web3+Leaders+Roundtable+Report.pdf/ca58a56837e511f1956882b60804cfcb
FAQ
Who published the Web3 Leaders Roundtable report?
The report is a joint publication by the Emerging Technologies function of the ADGM Registration Authority and Hashed, presenting the proceedings of a closed-door roundtable convened at Abu Dhabi Finance Week 2025. It was led by Dmitry Fedotov of ADGM and Simon Kim, Chief Executive Officer of Hashed.
What are the three core findings of the report?
First, financial infrastructure is likely to evolve toward machine-native execution and settlement to support autonomous AI agents. Second, governance and evidentiary integrity — including tamper-evident records of data, model versions, and agent actions — are likely to become binding requirements for AI systems. Third, tokenisation has entered a market-structure phase in which settlement, collateral mobility, identity, and prudential treatment determine scalability rather than issuance mechanics.
Which regulatory frictions did participants identify as most binding?
Participants cited Basel prudential capital treatment — including risk-weighted asset charges of up to 1,250% on some digital-asset exposures — fragmented FATF travel-rule implementation, inconsistent custody expectations across jurisdictions, and unresolved accounting and tax treatment as compounding constraints on institutional deployment, even where licensing pathways and market infrastructure already exist.
The report stops short of regulatory recommendations, framing its observations as “independent regulatory analysis informed by the roundtable dialogue.” Its concluding observation places the burden squarely on policymakers and market participants jointly: ensuring governance frameworks evolve in a way that preserves institutional safeguards without reproducing legacy frictions on new infrastructure. Whether the next phase of supervisory work — particularly on identity portability, settlement-asset treatment for stablecoins, and prudential treatment of tokenised exposures — closes that gap will set the trajectory for institutional adoption through 2026 and beyond.
