xxx

As AI agents extract only the data they need and move on, the banner advertising model is breaking down, and platforms are shifting toward charging agents directly.
To handle the vast number of automated micropayments agents generate, fee-free stablecoins and on-chain payment standards such as x402 are emerging as essential infrastructure.
As happened with payments in the music streaming market, the per-transaction price will fall, but the sheer volume of machine-driven usage will outweigh that decline, making the overall market substantially larger.

From: AI Agent Payments: The Inevitable Future.

xxx

Sebastien Taveau, then the CTO of Validity Inc., was kind enough to quote me in his review of 2012:

He coined the following statement that has become my favorite of the year.

“Identity is the new money”.

It is simple, powerful and summarizes exactly where the ecosystem is going.

[From

Looking Back Forward | Validity Inc. | Biometric Sensors for Mobile Devices

]

Seb is much too kind. I may well be guilty of popularising the aphorism in the context of payments and organisational strategies towards secure electronic transactions in the retail space, but I didn’t invent it. I heard it for the first time a few years ago in connection with the ill-fated UK national identity card scheme. I was at the time a member of the Home Office’s Advisory Forum and was interviewed by Sir James Crobsy, who had been called in by the then-Chancellor Gordon Brown to prepare a report on the scheme. It was Sir James who brought the phrase to my attention.

If, as Sir James Crosby said in his report on the U.K. ID card scheme, “identity is the new money”, then banks should already have generated strategic plans to accumulate the former, now that they’ve run out of the latter.

[From

Digital Identity: I’m sure banks have a strategy for this kind of thing

]

As time has gone by, I have become more convinced that there is a deep truth in the apparently simple statement and I’d like to explain why. But to do that, we have to first explore what money means. One of the problems that always comes up when discussing money is that the word means several different things. I want to focus on just two here: money as a generalised means of exchange between buyer and seller and money the subset of means of exchange that do not involve credit. In other word, cash. Identity changes the requirements for and use of both kinds of money.

If you know who all of the counterparties to a transaction are, and can establish their “credit” then there is no need for cash. Identity substitutes for cash: when I go into Waitrose and pay with my John Lewis MasterCard, it’s an identity transaction. The terminal in Waitrose establishes that I have access to a line of credit that means that Waitrose will be paid. No actual money moves between my card and the Waitrose till. On the other hand, when I buy an apple from a market stall and pay for it with a pound coin, the stallholder doesn’t need to waste any time or money trying to establish who I am, because he doesn’t need to trust me. He just needs to trust the pound coin, which he self-assays. It’s not that there are no counterfeit pound coins, because there are, but that there are too few of them to disrupt commerce (and, to be honest, if you give the smallholder a counterfeit coin and he later detects the fraud, he will probably just palm it off on someone else).

As a thought experiment, then, imagine that cash vanishes and we interact through identity. In that case, identity becomes the key to transactions and a crucial individual resource that needs to be looked after by responsible organisations. This is the idea behind the Digital Asset Grid put forward by the Innotribe team at SWIFT, the worldwide interbank messaging service, at last year’s SIBOS. Whether you think DAG is the right specific approach or not, there’s something to be said for begin strategic planning around the transition to identity-based transactions.

What does all this mean at a macro level? It means that the action in the payments world will shift further toward identity over the coming year. One of the reasons why the Single European Payment Area (SEPA) hasn’t transformed cross-border commerce in the way that had been hoped is that a great deal of cross-border commerce rests on identity, which is undoubtedly why the Commission has switched its attention and proposed new rules to enable cross-border and secure electronic transactions in Europe.

The proposed Regulation will ensure people and businesses can use their own national electronic identification schemes (e-IDs) to access public services in other EU countries where e-IDs are available. It also creates an internal market for e-Signatures and related online trust services across borders, by ensuring these services will work across borders and have the same legal status as traditional paper based processes.

[From

EUROPA – Press Releases – Digital Agenda: new Regulation to enable cross-border electronic signatures and to get more value out of electronic identification in Digital Single Market

]

You can see where they are coming from. The UK, however, does not have a national e-ID and is unlikely to have one for the foreseeable future. We’ve taken another path, using a framework approach and private sector identities, so a pan-European solution will have to work with public and private sector identities in a single framework. This line of thinking suggests that a fruitful line of enquiry might be to look into a pan-European trust framework that these identities can belong to.

In digital identity systems, a trust framework is a certification program that enables a party who accepts a digital identity credential (called the relying party) to trust the identity, security, and privacy policies of the party who issues the credential (called the identity service provider) and vice versa

[From

What is a Trust Framework? | Open Identity Exchange

]

Let’s hope that the Commission can help something like this to develop, because the real barrier to cross-border trade within the Single Market is not money, but identity.

Earlier this year, Sony got conditional approval from the OCC for a national trust bank in the US for the express purpose of issuing and managing a dollar stablecoin, launching next year. 

xxx

Financial institutions filed 15,417 reports of mail theft-related check fraud over six months in 2023, covering more than $688 million in suspicious activity, the analysis found.

The median report was $14,215.

From: Postal carriers charged in $24M Houston check theft | American Banker.

xxx

xxx

For all the interest in and uptake of agentic AI systems over the past year or so, the world of automated AI has thus far been primarily limited to text, images, code, and other data and actions that take place inside a computer. Anthropic is now aiming to change that somewhat with what it’s calling the Model Hardware Standard (MHS), a set of standardized drivers designed to let AI agents easily interface with and control arbitrary devices.

From: Anthropic’s new hardware standard lets AI agents control the physical world – Ars Technica.

xxx

It seems as if every day we see reports of “smart” (they are not smart) “contracts” (they are not contracts) being hacked, misused and abused for nefarious purposes. This set me wondering. How does the plundering of not-smart not-contracts in the emerging token economy compare to the plundering of fiat currency from banks in the old economy? 

When I say not-smart not-contracts are being subverted on a daily basis I am not by any means exaggerating. Here are two such events, chosen at random from a single day when I was writing something about risk recently: Fogo and Avici. An attacker subverted the Fogo token contract to siphon off 4% of Fogo’s “genesis” supply and more than 10% of current circulating supply while the Solana-based crypto card platform Avici, a self-custodial wallet connected to a secured Visa card, suffered a security breach where an attacker drained more than $1 million from user collateral accounts (while using only about $190 in initial capital). The exploit sent the AVICI token down roughly 49% in 24 hours.

Not all are smart contract raids

xxx

Blockstream’s Liquid Network, one of the oldest and most widely used Bitcoin sidechains, lost roughly 4,000 BTC — worth about $320 million — after attackers exploited a software flaw on September 6. The Liquid Network hack drained the federation’s reserves from more than 4,200 BTC down to just around 197 BTC in a matter of hours, instantly ranking among the biggest security incidents to hit Bitcoin-adjacent infrastructure this year.

From: Liquid Network Hack Drains 4,000 BTC in Major Security Breach.

xxx

the flaw sat upstream, in the software that validates transactions before they reach the 11-of-15 signing federation.

Is this crypto-native LLM-powered larceny a genuine replacement for old-school bank robbery or is it just grabbing the headlines in my feeds? Rather than do any useful work, I decided to find some figures. Given the romanticised stories of Bonnie and Clyde and the like, I turned to Depression-era America for the baseline statistic. It turns out that only around one or two bank branches per day were robbed during 1932, the first year that has vaguely reliable statistics.

That didn’t seem too bad to me, given the size of the United States, the number of bank branches and the number of guns and actually, it wasn’t that bad. In fact, when it comes to bank robbery, the Depression was far, far below America’s peak year of 1991, when the FBI recorded 9,388 bank robberies. Since then, the number has fallen considerably, presumably because there are fewer bank branches, because branches have less cash in them and because branch security has continued to improve.

Even so, there were still 1,788 bank robberies in the US in 2020 and I could not help but note that by comparison, in largely cash-free Sweden, there were five bank robberies in 2020 (and only three of these were armed robberies), down from the the 2011 peak of 43.

Rather than rob banks, criminals with a need for cash have shifted to target places outside vaults where cash is more vulnerable. ATMs, for example. There is a plague of ATM robberies in many countries, where criminals will either blow up ATMs or wreck buildings to carry off ATMs whole. Cash-in-Transit (CIT) is another target for cash-hungry criminals and the robberies can be very violent.

(At a conference I attended recently, one presenter showed some particularly graphic security camera footage of armoured cars being blown up on busy streets by criminals with no thought for the lives of the crews or passers-by.)

Bank robbery is a poor career choice. Franky, if you think that robbing banks is the way to a comfortable life then you would be better off going to work for one. Apart from anything else, if you get caught you’ll do a lot less time. In England and Wales, official figures show that people receiving immediate custody for robbery received an average term of about 3 years and 4½ months in 2023, compared with about 1 year  and 10½ months for fraud overall.

(The disparity is probably larger for genuinely armed robbery, but official statistics do not isolate either armed robbery or, for that matter, fraud perpetrated specifically by bank employees.)

Blockchain robbery seems like a much better career path than bank robbery. Chainalysis report $3.2 billion stolen from the cryptoverse in 2025. DefiLlama has logged 233 separate incidents so far in 2026, worth roughly $1.31 billion. Over the last decade, DeFi projects and crypto exchanges are estimated to have lost more than $14 billion to hacks and exploits, a figure that is undoubtedly conservative as it does not account for individual wallet hacks (or the secondary damage caused by an erosion of trust across the ecosystem).

Decision Time

So: as an unscrupulous capitalist investor hoping to steal my way to early retirement, should I invest money in bank robbery or smart contract robbery? It’s a no brainer. The rewards for bank robbery are limited and the perpetrators have a high likelihood of getting caught. The FBI no longer records the losses, but in 2019 the average haul was only $4,000-$4,200. Meanwhile, 

I asked AI to put some figures to my suspicion and it told me that the average not-smart not-contract vulnerability incident in the Beosin/Alert 2025 dataset was around $9 million, which means by my calculations the typical 2025 blockchain exploit was more than 2,000 times the size of a typical physical bank robbery.

Overall, then, it looks like one of the vocations that will vanish from history under the onslaught of AI, alongside management consulting and hacking (Elon Musk says that AI will be better than people at hacking by the end of next year), is that of the bank robber.

Solana-based crypto card platform Avici, a self-custodial wallet connected to a secured Visa card, suffered a security breach where an attacker drained more than $1 million from user collateral accounts (while using only about $190 in initial capital). The exploit sent the AVICI token down roughly 49% in 24 hours.

xxx

None of this means the BoE has decided to launch a digital pound through the back door; it has said clearly it has not. It means the boundary between public and private money in the UK is being redrawn through a sequence of individually defensible technical decisions, not through any single decision to redraw it. That is worth watching closely as the Code of Practice consultation closes this September and the regime moves towards its planned 2027 launch.

The EU, in particular, should be paying attention, for a reason that goes beyond simple competition. Brussels is building two things at once: a digital euro deliberately capped at roughly €3,000 per person to protect bank deposit funding, and a stablecoin regime under the Markets in Crypto-Assets Regulation  that bans remuneration entirely, with no carve-out for activity-based rewards at all – leaving EU issuers in a grey zone

From: Britain’s synthetic CBDC – OMFIF.

xxx

xxx

One especially noteworthy part for monetary policy-makers: the BoE confirmed it will build a central bank liquidity facility for systemic stablecoin issuers, offering short-term collateralised loans against their gilt holdings. The BoE describes this carefully as a backstop, not a front-stop: available to issuers that are fundamentally solvent, not a substitute for the 1:1 backing requirement or a fix for insolvency. Design details will follow in 2027, with access for eligible firms shortly after.

A private, non-bank issuer of digital money that holds most of its reserves in gilts and has a standing line to the BoE’s own balance sheet is acquiring something close to the defining feature of public money: a credible promise, backed by the central bank, that it remains convertible at par under stress. Call it what it functionally is: a synthetic central bank digital currency, built through a sequence of individually reasonable technical decisions rather than any single choice to create one.

That matters because banks earn lender-of-last-resort access in exchange for capital requirements, liquidity ratios, resolution regimes and continuous supervision. Systemic stablecoin issuers are being offered a version of the backstop before comparably mature discipline exists alongside it. The consultation on the draft Code of Practice, open for feedback until 22 September, is where that gap gets tested, and where issuers have every incentive to keep the backstop while resisting the obligations that historically came with it.

From: Britain’s synthetic CBDC – OMFIF.

xxx

xxx

A former White House teleprompter operator has been ordered to pay more than $172,000 (£127,000) for using inside information to bet on Donald Trump’s speeches.
Gabriel Perez was found to have made trades concerning what the US president would say during addresses between December 2025 and February 2026 on Kalshi – a prediction markets platform where users can bet on real-world events.

From: Ex-White House teleprompter operator ordered to pay $172,000 for Trump speech bets.

xxx