Get the hottest Fintech Singapore News once a month in your Inbox
In its latest report,Tokenized Markets: Practical Implementation Pathways for Authorities Considering Tokenization, the Global Finance & Technology Network (GFTN) sets out eight frictions in tokenised markets or tokenised market challenges.
These frictions, identified through firms and research partners building tokenised markets, came to light during GFTNâs open call and working groups. GFTN shares that the problem statements are the views of contributors, and not GFTNâs.
The eight problem statements highlight a steady pattern:
- Every friction engages both compliance and governance, as well as technology
- Seven out of the eight frictions steer towards coordination being a barrier to shifting from pilot to production
- Settlement is the most contested market function, including which money settles and when settlement becomes legally final
Here are three key tokenised market challenges you should know about.
1. Stablecoin Settlement Within Regulated Payment Licences
Cross-border payments pass funds from one bank to another through correspondents, and the currency is converted as the payment moves across these bodies. However, the GFTN report shared a notable percentage on how stablecoins move the same value for a much lower cost and in less time:
âIn Banca dâItaliaâs test transfers, the on-chain leg took under 15 minutes in seven of eight corridors and cost 0.4% of the amount on average,â
the GFTN report shared.
âIn Banca dâItaliaâs test transfers, the on-chain leg took under 15 minutes in seven of eight corridors and cost 0.4% of the amount on average,â
the GFTN report shared.
Transfers like these would usually take a âstablecoin sandwich formâ, where a local currency is converted into a stablecoin, which moves on-chain before being converted into local currency on the other side.
Here, a licenced payment provider handles the steps while the customer never touches the stablecoin. The provider, meanwhile, holds the stablecoin for as long as the transfer takes.
Why a New Rail Means a New Licence
Adding a new card scheme or bank partner rarely requires a licenced payment provider to obtain a new licence. However, using stablecoins as a settlement rail tends to require that.
Most jurisdictions treat fiat payments and digital assets as separate regulated activities, often overseen by different authorities. Providers then need to secure digital asset authorisation, establish a separately licenced entity, or partner with a licenced firm where permitted.
In markets with no framework or where stablecoins are banned, the service is off the table entirely.
Treating Stablecoins Like Any Other Rail
The contributor proposes that, where a payment providerâs existing licence and controls already cover the activity and its risks, adding a stablecoin rail should be treated like adding a local payment method or new card scheme.
The provider would stay accountable for the risks, but switching rails would not in itself trigger a new licence or variation. Notification and ongoing supervision could suffice.
This would apply to both ends: to the sending provider and the receiving provider that converts the stablecoin and pays out in local currency.
In Singapore, the Payment Services Act 2019 regulates digital payment token services as a separate payment service, and licensees must obtain a licence variation before adding a new service.
2. Where Do Payment Rules Fall Short When AI Agents Pay?
Payment rules are typically formed around two parties that can start a payment: a person and a licenced institution that acts for them. But what happens when AI agents are neither, yet thrown into the mix?
This additional party brings in new questions. A payment system would have to answer many things, from who is paying to whether the agent really has the authority to pay and, critically, whether it stayed within it.
While the Tools and Agents Exist, the Rules for Relying on Them Donât
Tools that would enable an agent to pay within stipulated limits are currently working in prototypes, like multi-signature approvals and conditional locks, but still, the missing piece is the segment of rules that would let others rely on them.
Customers would typically switch to a new payment mode only if they know who to call if something goes awry, but this does not exist yet for an agent.
The harder case is an agent that holds its own wallet and has no regulated operator behind it.
Three Gaps Holding Agentic Payments Back
In an October 2026 roundtable featuring a central bank, law enforcement, international and development finance institutions as well as the industry, three gaps were identified:
- Authorisation, KYC, and AML obligations remain national while agents move value within seconds. As there is no shared basis for reliance, checks are repeated at every step, or each jurisdiction issues its own agent credentials
- While transactions settle within seconds, requests for assistance between jurisdictions can take months
- As agentic AI is an orchestration layer, economies still building their digital public infrastructure, open APIs, and consent protocols cannot skip these foundations on the way to readiness
MASâs Safeguards for Agentic Finance at Runtime, published in July 2026, covers three areas: how agent actions are authorised, when human oversight is triggered, and whatâs recorded at each decision.
China has since issued âknow your agentâ guidelines for agent-driven payments, while the UK has consulted on how its payment rules should accommodate agentic AI, the report shared.
Yet two questions remain open. Who bears the loss when an agent acts within its mandate but against the principalâs intent? And how does one jurisdiction recognise a mandate issued under anotherâs rules?
3. Why Tokenised Assets Need Quantum Protection Now
As tokenisation moves data and value across networks, the cryptography that secures a token also protects the asset itself. This cryptography includes public-key methods to sign transfers and secure the channels between ledgers.
Future quantum computers are expected to be able to break it. A quantum machine powerful enough to defeat widely used RSA and elliptic-curve methods is what cryptographers call âcryptanalytically relevant.â
Confidentiality Risk Comes First, Integrity Next
Confidentiality is the nearest risk to tokenised assets. Data secured by todayâs key exchange, like data passing between ledger environments, can be harvested now and decrypted later.
Integrity risk is further out, when quantum computers can forge signatures or recover private keys.
A Migration That Takes Years
Migration, meanwhile, takes years. Building post-quantum cryptography directly into a ledger can slow it, as post-quantum keys and signatures are larger, which can increase block size and slow signing time.
The task, then, is to protect tokenised assets now without rearchitecting the underlying ledgers, while preparing to migrate transaction signing before the threat arises.
Governments and regulators already treat migration as a current obligation. Here, the Monetary Authority of Singapore has issued an advisory to guide financial institutions in addressing the cybersecurity risks posed by quantum computing.
As GFTN notes, the problem statements come from parties with a stake in the outcome and are presented as such. They are not conclusions to adopt as given. Instead, each country will need to weigh them against its own starting conditions.
The answers may change from one market to another, but the questions are certainly not to be ignored.
Featured image edited by Fintech News Singapore based on an image by magnific on Magnific



