Why the AI era needs a society, not just a wallet
The account protects the person. The aggregate protects the people.
A living room in Manchester
Manchester, New Hampshire, was a textile city in 1908. The Amoskeag mills employed a large workforce, including French Canadian families whose lives were organized around demanding work and narrow financial margins.
The work was waiting. So was something else. The wages were low, the rents were high, and the workers had no real access to the city’s banks, which were not interested in small deposits from immigrants who could not always sign their names in English, and not interested in small loans to families who needed forty dollars to bury a parent or one hundred to bring a sister down from Trois-Rivières. What the banks were not interested in, the loan sharks were. Predatory lending was the dominant form of credit available to the Amoskeag workers, and the cycles of debt that followed were ordinary enough to be a feature of parish life.
Monsignor Pierre Hévey, pastor of Sainte-Marie’s parish, sought a different arrangement for his community. Alphonse Desjardins had helped establish cooperative savings and lending in Quebec. The idea was that people could organize their own institution around a common bond and a pool of savings.
Hévey wrote to Desjardins in the spring of 1908 and asked him to come down. Desjardins came. The two of them, with a Manchester attorney named Joseph Boivin, organized the first United States credit union in the parlor of Boivin’s home on Notre Dame Avenue. It opened on November 24, 1908. The first members were Amoskeag workers and their families; a share cost five dollars; deposits were taken from workers, families, and children. For the first several years the work was conducted entirely in Boivin’s living room, the safe a metal box bought used from L’Avenir National, the city’s French daily, the staff a group of volunteers with Boivin himself unpaid as president and manager.
St. Mary’s received its New Hampshire charter in 1909. Massachusetts enacted a general credit union law that year, and the Federal Credit Union Act followed in 1934. A local practice became part of a much larger institutional and regulatory framework.
By December 31, 2025, the National Credit Union Administration reported 4,287 federally insured credit unions, with 144.7 million memberships and $2.43 trillion in assets. Memberships are not a count of distinct people: one person can belong to more than one institution. The figures establish substantial scale without making the form universal.
It has lasted for the reason any institution lasts when it lasts: it was built to do something ordinary commerce was not going to do, and built well enough that the doing of it could outlive its founders. The argument of this chapter is that the AI era is now asking for the same institutional move, applied to a new substrate, and that the credit union is the second ancestor of Cyberwealths because it knows how to make the move. An account is not enough. The AI era is being offered a great many containers in which a person can hold what their life produces, and very few societies in which a people can hold it together. This chapter is about the difference.
What a credit union is, and what its discipline costs
A credit union, in its plainest formulation, is a nonprofit financial institution owned by its members and governed for their benefit. Each member, on joining, buys a share — typically a small amount, five dollars at St. Mary’s a century ago and five dollars at St. Mary’s today, and the share establishes membership. Members deposit savings into share accounts; the institution lends those deposits to other members on terms set by an elected board and consistent with its mission; it pays interest on deposits, charges interest on loans, and uses the spread to cover costs, build reserves, and return surplus to members through better rates, lower fees, dividends, and community services. Governance is one-member-one-vote, regardless of the size of a member’s deposit. There are no outside shareholders.
What this structure refuses is the trade by which a financial institution exists primarily to maximize returns to investors who are not its customers. That refusal is the heart of the form. A commercial bank is organized around outside capital: shareholders provide equity, the bank generates returns by lending to and charging fees on its customers, and the customers’ interests are managed against the shareholders’ in an arrangement that, however well-regulated, leaves no doubt about whose returns the institution is structured to maximize. A credit union refuses that arrangement at the level of corporate form. The members are both the customers and the owners. There is no outside party whose returns have to be served against them. Its essential discipline is that the alternative to investor ownership is neither state ownership nor customer-management but member ownership under a common bond — a shared workplace, parish, region, profession, or school.
Credit unions can offer products similar to banks, but their member ownership changes the institutional purpose. Surplus can support reserves, services, rates, and other member benefits. Growth, mergers, capital constraints, and management choices still matter. Member ownership supplies a discipline; it does not remove ordinary business decisions or conflicts.
Members still need competent management, sufficient reserves, and effective oversight. A cooperative may decline an opportunity inconsistent with its purpose, but slower growth or inferior rates are not inherent requirements. The institution should be judged by how well it serves members while meeting its obligations.
Mutuality does not confer immunity from financial crises. Some credit unions suffered serious losses during the 2008 crisis, and the corporate credit union sector required substantial intervention. The lesson is to combine member purpose with sound risk management, supervision, and credible protection for members—not to assume the ownership form makes failure impossible.
The wallet is a container, but it is not a society
The AI era is being offered a substitute for the credit union — the individual digital wallet — and the substitute is inadequate to the work the form has to do. The wallet has its appeal. In a world where authored experience, learning records, AI representative authority, credentials, consent receipts, and the structured representations of personal life are about to become foundational economic assets, the question of where a person holds those assets is real and urgent, and the wallet appears to answer it. Put credentials in a wallet, tokens in a wallet, keys in a wallet, identity in a wallet, the AI representative’s authorization in a wallet. The person becomes sovereign over the assets, the wallet is the instrument of that sovereignty, and the architecture is clean.
The architecture is also incomplete. A wallet is a container. It holds what you put in it. It does not, on its own, bargain for you, govern on your behalf, pool your risk with anyone else’s, finance shared infrastructure, set common standards, negotiate model-training terms, or distribute the value that the use of your authored experience helps produce. A wallet, on its own, is a single person facing an unequal contracting environment with no institutional counterweight, and the counterparty is the platforms, the cloud providers, the model developers, the data brokers, the AI service companies, and the owners of the spatial computing layer that is about to render most of public life. The single person, against all of that, has a wallet. The wallet is real. The asymmetry it leaves the person in is also real.
The wallet is a container. The credit union is a society.
What St. Mary’s understood in 1908 is that a single Amoskeag worker with savings in his pocket was not actually safer than that same worker with savings in a society. The pocket was still vulnerable to theft, to fire, to misjudgment, to the small emergencies that drain individual savings without warning. More importantly, the pocket could not become a loan. The worker who needed to buy a small house, bring a sister down from Trois-Rivières, open a corner grocery, or pay a doctor could not turn his own pocket into credit. The institution did that work, and it did it because it was a society, not a container. The same is true of authored memory, AI representative authority, and the rest of the assets that will define the Cyberwealths of the AI era. A person can hold these in a wallet. The person cannot, on the strength of the wallet alone, bargain over their use, finance shared infrastructure with them, pool them into models that answer to their owners, set terms for their training, or build the institutional counterweight to platforms that the era requires. The wallet is a container. The institution is the work.
This is not a critique of wallets as tools. Cryptographic accounts, key management, signed permissions, attested credentials, and portable identity primitives are necessary infrastructure for any AI-era institution — necessary precisely because they let an institution operate without having to centralize the kind of trust that platforms now centralize through ownership. The argument is that the tools, by themselves, are not the institution. They are the substrate on which the institution has to be built. The credit union form, applied to authored experience and AI representation, is the institution.
A Mutual AI Credit Union is not a DAO and is not an exchange
Two adjacent confusions are worth refusing directly, because the AI-era discourse has already produced them and each occupies institutional ground the Mutual AI Credit Union needs to claim.
A DAO (pronounced "dow") stands for a Decentralized Autonomous Organization. It describes a broad family of organizational experiments in using ledger-based technologies to organizing without central authority. Many use token-weighted voting, which can concentrate authority among large holders; others use different membership or voting arrangements. Code-based coordination may be useful, but a member-governed AI institution should choose its rules explicitly rather than assume decentralization guarantees equal voice.
A cryptocurrency exchange is primarily a trading venue. Its ownership, custody, automation, and governance vary, and it need not operate through smart contracts. Trading infrastructure alone does not provide the common purpose, member control, or obligations to participants that the credit-union analogy is meant to emphasize.
The distinction I want to preserve is member standing: understandable rights and obligations that do not rise and fall simply with a tradable token’s price. Different technical arrangements could support that purpose. Their governance should be assessed directly rather than inferred from the label DAO, exchange, or cooperative.
The Green Bay Packers offer a familiar example of ownership organized around a shared purpose. In a September 2026 New York Times essay, Robert J. Geline describes a team whose community attachment reaches far beyond its home city. The Packers have operated as a publicly owned nonprofit corporation since 1923. Fans have supplied capital through successive stock sales; shareholders receive no dividends, and a board of directors and executive committee govern the organization.
People help hold an institution they want to remain part of their lives. That is a different return from a financial distribution, and both forms of value matter to the ecosystem envisioned here. The Packers also participate in the NFL’s national revenue system alongside their local business. The example suggests how an institution can retain a particular home and community identity while drawing strength from a much larger network.
The data backpack is the AI era’s share account
If the credit union form is what the Cyberwealths era needs, the first practical question is what the AI-era equivalent of the share account looks like. The share account is the institutional anchor of credit union membership: a member opens it, deposits the initial five dollars or its equivalent, and from that moment is a member with rights, obligations, and standing inside the institution. Everything else the member does — savings, loans, governance, dividends — flows from it.
The AI-era equivalent, and the term that has been developing across the Into Spatia work and elsewhere, is the data backpack. It is a secure, member-owned, portable account holding the structured representations of authored experience that the AI era is going to make economically and personally consequential — not a wallet of tokens but the member account of a Mutual AI Credit Union, the AI-era analogue of what St. Mary’s offered the Amoskeag workers in 1908. What it holds, in operational terms: the authored experience the member has chosen to record, in structured forms that preserve provenance and consent; the credentials, attestations, and consent receipts that govern who may use what under what terms; the relationships the member’s AI representative has built across time, including preferences, operational rules, boundaries, and audit records of past actions; the learning records that document what the member has done, made, and become, structured for the member’s own use rather than for platform profiling; the dividend entitlements accumulated through participation in shared infrastructure; and the membership rights, governance roles, and standing the member holds inside the institution.
Unlike money in a financial account, these records retain the context of a person’s life and relationships. The institution needs rules for copying, derivative uses, and revised or withdrawn permissions, including choices made when a member was younger or differently situated.
The list is operational rather than exhaustive. The point is that the data backpack anchors membership in the AI era the way the share account anchors it at a credit union. It is where the member’s relationship to the institution is concentrated, recorded, and made governable, and it is also where the member’s relationship to other members is mediated, because the backpack is what enables the pooling, the governance, and the dividend distribution that are the institution’s reason for existing. It is portable but not lonely. It belongs to the person but is supported by an institution with duties to the membership as a whole. It is technically secure and socially governed.
A member account would sit within collectively agreed rules, but collective governance should not silently override the permissions attached to personal records. The institution needs to distinguish decisions members make together from uses that require an individual’s authorization. It must also explain what withdrawal can and cannot reverse, especially after information has been disclosed or used in training.
The dividend is the return of value, not the price of a person
One of the more persistent confusions in the discourse around personal data is the assumption that the alternative to extraction is selling. If platforms are extracting value from authored experience without compensation, the assumption runs, then the answer must be to pay the person for the use of their data. The framing is wrong in a way that matters, and the credit union gets it right.
The relevant distinction is between a negotiated payment for a specific use and a continuing claim on an institution’s benefits. A cooperative could offer both. The design question is whether members receive useful services, an accountable voice, and agreed economic benefits, with clear terms for each.
Collective returns can broaden the benefit beyond a separate price for each person’s record. They do not eliminate pressure to commercialize information. A member institution still needs clear limits on eligible uses, consent, conflicts of interest, and distributions, including a meaningful option not to contribute private material.
Benefits might include lower fees, better services, shared infrastructure, community investment, or distributions where appropriate. Equal voting rights should remain equal. A “governance dividend” should mean greater practical capacity to exercise a member’s voice, not additional votes for those whose data or income is valued more highly.
Why the aggregate is the heart of the form
The core argument of this chapter, and the reason the credit union is the second ancestor of Cyberwealths rather than the first or the third, is that the aggregate is what an individual account cannot be alone. Every credit union begins from the same premise: small accounts, pooled, can do what no small account can do alone. A worker’s ten-cent deposit at the original caisse in Lévis was, by itself, ten cents. The same ten cents, pooled with the deposits of a thousand other workers, became the capital base for loans that could buy a house, start a business, get a sick child to a doctor, or carry a family through a layoff. The capital base existed because the deposits existed; the deposits existed because the institution existed; the institution existed because the members had built it. The architecture was circular by design, and the circle was what produced the power.
The same circular architecture is what a Mutual AI Credit Union has to produce. A single person’s authored experience, held in a single data backpack, is by itself an asset of limited consequence. Pooled with the authored experience of a thousand other members, it becomes a basis for what no single member could do alone: negotiating model-training terms with platforms that no individual could secure; funding cooperative compute, storage, and security infrastructure that no individual could afford; setting common consent standards that platforms have to meet to interact with the membership at all; developing and governing shared AI representative models that answer to member-set rules rather than platform incentives; refusing extractive uses collectively in ways no individual refusal could effect; and pooling the legal, regulatory, and technical capacity that no individual member could develop alone.
Federation is how this ecosystem can scale. Local institutions can share compute, technical services, purchasing, and bargaining capacity while members retain authority over their own institution and records. Each new community adds capacity to the network. Shared services need their own accountable governance, with clear responsibilities, costs, and routes for withdrawal. The aim is to make cooperation between communities strong enough to support services none could sustain alone, while keeping decisions answerable to the people affected by them.
This is the move the chapter has been building toward. The wallet is not the answer because the wallet does not aggregate. The DAO is not the answer because it aggregates tokens rather than members. The exchange is not the answer because it aggregates transactions rather than obligations. The Mutual AI Credit Union is the answer because it aggregates members — persons with shares, votes, standing, obligations to one another, and a common bond strong enough to support the trust that aggregation requires. The aggregate is also what lets the institution act at the timescales the AI era demands. Individual consent is too slow for AI-era decision-making and too poorly informed for AI-era complexity; institutional governance, with elected members, fiduciary duties, and the discipline of a common bond, can act at those timescales and with more information and accountability than any individual could bring to bear. The account protects the person. The aggregate protects the people.
What the Mutual AI Credit Union inherits from St. Mary’s
Mutual AI Credit Union is an analogy and a working name in this book, not a claim that the proposed organization is a chartered financial credit union or that its accounts would be insured deposits. The inheritance is member purpose, mutual benefit, and accountable governance. A real organization must identify its activities and obligations, select a lawful structure, and describe its services and protections accurately.
What the common bond looks like in the AI era is one of the chapter’s open questions. In the early credit union era it was usually a single employer, parish, or town; in the modern era it has broadened to metropolitan regions and entire industries. The Mutual AI Credit Union will need bonds adequate to its purposes — a school district whose students and families maintain AI representatives and data backpacks together; a city or region developing a local knowledge graph and a spatial computing layer; a profession needing bargaining capacity in AI-mediated employment; a cultural, religious, or mutual aid community preserving memory and practice across the AI transition; an aligned AI Trust whose members agree to shared rules for representation, consent, and benefit. The common bond is what makes the institution legitimate, and also what makes it functional, because trust within a defined bond can support governance at scales an anonymous market cannot.
The tradition arrives in stages, and each stage leaves a lesson the AI era needs. Nineteenth-century mutual aid and cooperative finance established that people pool risk and resources when ordinary commerce fails them; a mutual AI institution pools rights, compute, memory protections, and learning infrastructure for exactly that reason. The early credit unions of the 1900s and 1920s showed that safe accounts and fair credit could be built around a common bond; the data backpack account, governed through membership rather than platform terms, is the same claim about a different asset. The Federal Credit Union Act of 1934 established that mutual institutions could become durable public-interest infrastructure with standing in law, which is what AI Trusts will need, in the form of legal recognition, auditability, and accountability that outlasts their founders. The modern era demonstrated that surplus can return to members as better terms, dividends, and community investment; AI-era dividends would return as data, service, governance influence, and community benefit. And federation showed that local institutions can gain scale without surrendering local governance, which is how Mutual AI Credit Unions would bargain — regionally, nationally, or by profession — with counterparts operating at platform scale.
Cyberwealths begin where individual control becomes collective capacity.
The architecture is not unusual; a credit union does versions of all of it every day. The Mutual AI Credit Union applies the same architecture to a new substrate — harder in some ways, because the assets are newer, the legal vocabulary less developed, and the digital infrastructure still being built; easier in others, because the cooperative civic culture across rural and small-town America still understands the form, the legal infrastructure for cooperative ownership is well-developed, and the political case has the unusual property of being makeable convincingly across the ideological spectrum. The Mutual AI Credit Union sits between the Land Trust and the Employee-Owned Enterprise in the architecture of Cyberwealths: the Land Trust holds the foundational assets in trust, the Mutual AI Credit Union pools the means by which those assets become a working economy, and the Employee-Owned Enterprise organizes the productive work that the working economy supports. Each refuses a specific trade ordinary commerce would otherwise license. Together they make the AI era’s foundational institutions answerable to the people whose lives the era is being built around.
The institution beyond the founding room
The significance of the founding room is what people built beyond it: an institution that could serve members across changes in the local economy. Its history offers a discipline of continuity, not a promise that every mutual organization will endure.
There is one turn in this story I didn’t find in the archives. I found it by having worked in those buildings myself. In the early 2000s I served as director of operations for FIRST, the youth robotics organization that exists to draw young people into science and engineering, whose headquarters occupy the renovated Amoskeag mills along the Merrimack — the same millyard where the French-Canadian weavers had worked, whose hardship Hévey and Boivin had organized the credit union to answer. We were doing, for a new century’s young people, a version of what that millyard had always done: gathering people around the work of building things and sending them out able to build more. The buildings outlasted the company that raised them, exactly as the credit union outlasted the industry its members served, and they had become a place where the next generation of builders is made. I didn’t set out to write a chapter that ended where I had once reported to work. The form has a way of putting you back inside it.
The first one is still operating in Manchester. The form has crossed wars, depressions, financial crises, regulatory reorganizations, the rise and fall of industries, and the slow displacement of nearly every founding common bond by newer ones. None of the men in Joseph Boivin’s living room in 1908 could have anticipated any of it. They were doing one small piece of work for a population that needed it. The form did the rest.
The AI-era task is to adapt that discipline to records and services with different risks. A memory is not a deposit, and a Rep is not a loan. A useful institution would nevertheless have to answer familiar questions: who belongs, who decides, who benefits, and who is responsible when something goes wrong?
Sources
National Credit Union Administration. Historical Timeline (ncua.gov/about/historical-timeline). The standing federal reference for the chronology of the American credit union movement, including the 1908 opening of St. Mary’s, the 1909 New Hampshire charter and Massachusetts Credit Union Act, and the 1934 Federal Credit Union Act.
St. Mary’s Bank. Our History (stmarysbank.com/about-us/who-we-are/our-history). Primary source on the founding and continuing operation of St. Mary’s Cooperative Credit Association, including the roles of Monsignor Pierre Hévey, Joseph Boivin, and Alphonse Desjardins, and the original five-dollar share.
America’s Credit Union Museum, located in Joseph Boivin’s former home at 418–420 Notre Dame Avenue, Manchester, New Hampshire. Operating historical site documenting the early operation of the first United States credit union.
MyCreditUnion.gov. What Is a Credit Union? Standing reference on the not-for-profit, member-owned, cooperative structure of credit unions, including the principle that surplus returns to members through reduced fees, higher savings rates, and lower loan rates.
World Council of Credit Unions. History (woccu.org/about/history). Reference on the global expansion of the credit union form, including the 19th-century German cooperative-banking tradition (Schulze-Delitzsch, Raiffeisen) and Desjardins’s adaptation in Quebec.
National Credit Union Administration, Quarterly Credit Union Data Summary, 2025 Q4. Figures are for federally insured credit unions as of December 31, 2025: https://ncua.gov/files/publications/analysis/quarterly-data-summary-2025-Q4.pdf
New England Historical Society. First U.S. Credit Union Opens in Manchester, N.H. Background source on the 1908 founding, the Amoskeag textile-mill context, the role of Monsignor Hévey, and the assistance of Alphonse Desjardins.
National Credit Union Administration, 2008–2009 Annual Report, including the corporate credit union crisis and stabilization measures: https://ncua.gov/files/annual-reports/AR2008-2009.pdf
Robert J. Geline, “The Billionaire-Free Packers Stay True to the Game,” The New York Times, guest essay, September 20, 2026: https://www.nytimes.com/2026/09/20/opinion/green-bay-packers-nfl-owners.html
Green Bay Packers, “Green Bay Packers Shareholders,” stock-sale history and governance information, consulted September 2026: https://www.packers.com/community/shareholders
Mike Spofford, “Packers’ finances staying in good shape,” Green Bay Packers, July 23, 2025. Describes local business and national NFL revenue: https://www.packers.com/news/packers-finances-staying-in-good-shape-2025