A library of snippets
xxx
AI systems should be architecturally constrained to their intended purposes. Access controls, audit logging, and technical limitations should prevent unauthorized use cases. Systems should be designed so that misuse is difficult, not just prohibited.
From: When AI Meets Identity | Digital ID & Authentication Council of Canada.
xxx
xxx
The evidence of substantial, ai-fuelled productivity gains, however, is thin.
xxx
xxx
Key insight: As funding mechanisms for sovereign debt and large assets increasingly migrate toward real-time financial infrastructure, banks are at risk of being displaced as key intermediaries in global finance.
What’s at stake: Much of the infrastructure supporting capital markets was built for a paper-based environment and later digitized in layers rather than rebuilt. The result is a system that remains costly, slow and operationally intensive.
Forward look: The greatest strategic risk for banks is the quiet movement of real-world capital onto infrastructure that no longer requires manual intermediation.
For several years, much of the banking industry’s attention around digital assets has focused on volatility, regulation and reputational risk. The instability of crypto markets made caution understandable.But the structural shift now underway has little to do with speculative tokens.
Governments, sovereign funds and large asset owners are beginning to move real-world capital, including infrastructure, energy projects and national assets, onto real-time financial infrastructure that automates issuance, settlement and reporting. As these systems remove reconciliation, intermediaries and settlement delays, parts of the traditional banking value chain risk being designed out of the process.
This is not a new asset class. It is a new operating model for capital markets.
Banks have long played a central role in sovereign and large-scale project finance. They originate transactions, provide custody, manage settlement, reconcile positions and distribute products to investors.
Instant payments don’t fail. The systems around them do.
Real-time rails are only as reliable as the batch processes, fraud checks and integrations behind them.
PARTNER INSIGHTS FROM BMC
The economics of this model depend on operational complexity.Cross-border transactions still pass through multiple intermediaries and settlement cycles that can take days. Much of the infrastructure supporting capital markets was built for a paper-based environment and later digitized in layers rather than rebuilt. The result is a system that remains costly, slow and operationally intensive.
At the same time, the funding environment has become more challenging. McKinsey estimates that global infrastructure will require more than $100 trillion in cumulative investment by 2040, highlighting the scale of capital sovereign issuers must attract in the coming decades. Higher interest rates, tighter liquidity and rising sovereign financing needs are forcing governments and large asset owners to look for more efficient ways to access international capital.
Increasingly, the question is no longer how to digitize existing processes. The question is how to eliminate them.
A new generation of institutional financial infrastructure is beginning to support this shift. These systems automate the full lifecycle of financial assets, from issuance and ownership verification to settlement and reporting, on shared permissioned networks.
Settlement can occur in near real time rather than days later. Reconciliation across multiple institutions is replaced by a single shared record. Compliance requirements and asset conditions can be embedded directly into the instrument.
The impact is not limited to speed. It changes the operating model by reducing intermediaries, lowering operational risk and significantly decreasing administrative costs.
This transition is already underway. The market for tokenized real-world assets has expanded rapidly, reaching tens of billions of dollars in recent years. Several industry forecasts project the market could exceed $100 billion by 2026 as institutional adoption accelerates, with longer-term estimates suggesting the sector could scale to $11 trillion by the end of the decade if deployment moves beyond pilot programs and into core financial infrastructure.
xxx
xxx
Flare researchers analyzed322 posts published by 35 unique actors in a fraud-focused chat group revealing a structured resale economy built around compromised airline and hotel loyalty accounts, with 3,007 total travel vendor mentions.
From: Going the Extra Mile: Travel Rewards Turn into Underground Currency..
British Airways were 17th on the list. I can’t say I’m surprised. Even the most sophisticated AI enhanced criminal masterminds would find it a problem beyond the bounds of a battalion of LLMs to find an Avios seat on a fight to anywhere other than Dusseldorf.
xxx
NVIDIA and Microsoft are not alone in seeking to re-architect modern computing infrastructure to position LLMs and transformer-based technologies as the first point of contact between humans and information systems. Meta, owner of Facebook, has released research on how to extract behavioral data that signals intent from visual images. One paper introduces “Intentonomy,” a data set for human intent understanding (Jia et al., 2021).
xxx
xxx
Lloyds is set to stop opening accounts for customers in its branches, instead requiring them to use the bank’s app or website.
From: Lloyds stops opening some bank accounts for customers in branches.
xxx
xxx
Joining-up a fragmented public sector
Each department and each of its many separate, silo transactions essentially wants you to do the same thing. Scan your passport, provide proof of address, provide your national insurance number and other personal and contact details. Get something wrong, and you’re back to square one. Or worse, you miss an important deadline and get a fine.Government has spent nearly three decades failing to integrate and de-duplicate its systems, choosing instead to take the easy path — placing a thin digital veneer over the front-end of government. The lack of a sustained focus on the internal re-design and re-engineering of government has left it prioritising organisational needs over those of citizens and businesses.
Agentic AI potentially changes everything, despite still working from the outside-in. In fact, you could argue it’s almost as if agents were designed precisely to overcome the website-centric, organisation-focused divisions that characterise “digital government”.
The big difference is that now instead of suffering all the pain and frustration and humiliation of endlessly being asked the same thing, you can task your agent with the work. You open your personal AI assistant — such as Claude, or even maybe a new “Civic Agent” startup, or an enhanced Perplexity — and say: “I’m moving from Manchester to Lerwick on 15 April. Update my details everywhere and confirm it’s been done.”
xxx
It might be a good time to rethink what me mean by “financial literacty”
A study of 5,000 Britons commissioned by Lloyds Banking Group late last year concluded that more than half of adults were using AI platforms for financial advice.
As many as one in three were using the tools at least once a week for information or advice on money matters, with OpenAI’s ChatGPT the most-used platform, followed by Google’s Gemini.
Escalating adoption has also caused turmoil in markets, with wealth managers and brokers seeing sharp falls in their share prices last month after a US-based fintech launched an AI-led tool to help financial advisers personalise clients’ investment strategies.
xxx
xxx
There is another aspect to the relationship between stablecoins and CBDCs that I also find rather interesting. Earlier this year, the Governor of the Bank of England, Andrew Bailey, said in a speech to a G30 seminar in Washington that while commercial banks are the “best home” for innovation in retail CBDCs (I have to say that I am not clear why he thinks this), central banks should continue their preparations because “We have not yet seen enough evidence that the innovation will happen in commercial banks”.
It is clear that a crucial driver of the stablecoin boom is innovation: the people building next-generation cool stuff in their dorm rooms, particularly cool stuff around agentic finance, are using stablecoins to exchange value because stablecoins are a platform for innovation (which is what a CBDC should be). This widespread permissionless innovation is good for all of us.
I always thought that the innovation agenda was the most interesting element of the Bank of England’s analysis: Given Britain’s leading role in fintech, and the renewed commitment to the sector following the “Kalifa Review”, I saw the issue of competition as critical. A point well made in the Bank of England’s initial 2020 report was that the introduction of “smart money” (either by some form of smart contract usage or API interface) was really where the benefit of a CBDC would come from. CityUnited, a London think-tank, made that same point, arguing that digital Sterling must be “at the heart of Britain’s efforts” to strengthen London’s global role as a financial centre.
From: Central Banks Can Learn From Stablecoins How To Make Better CBDCs.
xxx
xxx
The mechanics behind this loss are brutal but standard. Decentralized exchanges (DEXs) rely on liquidity pools. When a buy order exceeds the available liquidity at the current price, the automated market maker (AMM) moves the price up the curve to fill the order.
To fill the $50M order, the protocol had to buy available AAVE at astronomically higher prices, resulting in an average entry price that wiped out the capital immediately.
This highlights why institutional players typically break such trades into thousands of smaller chunks or use OTC (over-the-counter) desks.
While Ethereum is quickly cementing itself as the backbone of institutional settlement, this event shows that the user interface layer still allows for catastrophic human error. Smart contracts do not judge the wisdom of a trade; it only executes the parameters signed by the wallet.
From: AAVE Crypto Swap Costs $50M as ETH MEV Pocketed $9.9M.
xxx
Snippings