Enforceable duties, not voluntary ethics
AI that works for you, not for them.
A 501(c)(3) nonprofit building embedded fiduciary governance for AI: agents bound to the person they serve, with the same duties of care and loyalty a doctor or lawyer owes a client.
Members of the GliaNet Alliance
What duty looks like, a series of fictional composite stories from the field.
On a Tuesday in March, an AI denied Marcus's daughter's autism therapy.
He appealed. Denied. He appealed again. Denied. This is where the old version of the story ends.
The narratives and characters are fictional, but the harms from these technologies are well documented — and occur every day.
About
Every institution that touches your lifehas put an AI at the front of it
Your health insurance
Uses AI to decide which claims to cover.
Your employer
Uses AI to screen resumes before a person even views them.
Your bank
Uses AI to set the rate on your loan.
Your child's school
Uses AI to score their writing.
Your supermarket
Uses AI modeling to decide what you will pay.
You stand on the other side of all of themwith no AI to protect or promote your best interest.
In every one of those encounters the AI is on the institution's side — paid by it, trained on its goals, loyal to its bottom line.
GliaNet exists to close that gap.
We are building the alternative: embedded fiduciary governance that has to live inside a system for it to be genuinely yours — and setting the standards that separates real accountability from simple 'trustworthy' marketing jargon.
You deserve a seat at the table — with an AI that answers only to you.
How it works on you
The dominant digital business model runs on a four-stage cycle. Each stage makes the next one easier.
- Surveil.
Your evenings, logged.
- Extract.
Your history, sold on.
- Analyze.
Your patterns, priced.
- Manipulate.
The price raised for you, not your neighbor.
— and what their AI learns from you feeds the next pass.
What the cycle means
Each stage reinforces the next — structurally incompatible with genuine accountability. Most governance efforts constrain the cycle from the outside: policies, audits, transparency reports. Responsibility gets articulated, but never assigned to anyone with authority inside the systems people actually use.
Where the work is happening
Almost all AI governance workson one side of the transaction.We work on the other.
The institution
An insurer, an employer, a bank, a school, a platform.
Its AI
Paid by it, trained on its goals, loyal to its bottom line.
the transaction
You
The person the decision is being made about.
Your AI
Nothing here. Today you arrive alone.
Supply-side governance
Rules for the companies that build and deploy AI — audits, transparency reports, safety standards, regulation. Necessary work. Almost everyone is doing it.
Demand-side governance
Building for the person the AI is pointed at — an agent bound to their interests, and the duties that make that binding real. This is the side GliaNet works on.
Every AI you meet works for someone else.We build the one that works for you.
Where we differ
AI governance is a crowded field, and most of it is doing necessary work: making the systems institutions point at people less harmful.
That work leaves one critical question unanswered — who builds for the person those systems are pointed at? Everything below follows from answering it.
Builds for the person on the other side of them. Nearly every AI governance effort works to constrain the systems institutions point at people. Necessary work — but it leaves the harder question unanswered: who builds for the individual?
Arrives carrying yours. Today you are the second party to every digital relationship — the passive recipient of someone else's terms of service. A fiduciary agent inverts that. It arrives as the first party, carrying your terms, your permissions, and your preferred ways of interacting. It negotiates rather than clicks accept.
Accepts duties someone else can hold it to. Care, fidelity and loyalty are legal duties carrying centuries of doctrine, established remedies and real liability. Responsible-AI principles are statements of intent — and the entity that issues them is also the entity that interprets them. A duty is something an entity can be held to by someone other than itself.
Sends someone to act for you. Privacy law shields people from harm. It never sends anyone to act for them. The missing middle is fidelity — the duty of faithful execution — which no competing framework supplies.
It's what the law is becoming. In May 2026 Utah made it real: companies handling your digital identity now owe you a duty of loyalty. Fiduciary duty is real law — real remedies, centuries of precedent, and not something a company can quietly revise. Whether it applies to digital systems is no longer an open question.
Not a different set of principles. A different side of the table.
From the Public Record
June 4–5, 2026 · Academic
Co-hosted "Fiduciary Duties and AI: Legal Frameworks, Technical Implementation, and Governance" with the Information Law Institute at NYU School of Law
Read moreJanuary 29–30, 2026 · Academic
Health Tech & Accessibility Roundtable, co-hosted with the Coleman Institute at CU Anschutz
Read more
What we do
GliaNet is building embedded fiduciary governance into
commercial, digital public infrastructure (DPI), and governmental AI agent systems.
The Pledge
We work with companies and organizations and help them take the first step towards a voluntary fiduciary commitment with their patrons.
Research
Through research, we translate fiduciary principles into governance models institutions can implement inside systems already in operation.
Consultation
Through consultation, we work alongside those institutions to embed the governance layers into daily practice.
Policy
Through policy engagement, we give lawmakers grounded input as the rules for AI agents are written.
Our mission is to build the accountability layer designed to keep systems loyally serving the best interest of the individuals and communities they touch. We work with partners seeking fully operationalized embedded fiduciary governance for existing or planned sandboxes across the healthcare, education, workplace, civic, marketplace, and smart-city domains. GliaNet has received initial philanthropic backing from Omidyar Network.
What do we do?
NOT
- A pledge alone
- A promise
- A pinky swear
NOT
- Harmful bots
- Disloyal avatars
- Double agents
NOT
- Big Tech
- Advertisers
- Fraudsters
We do Embedded Fiduciary Governance — For AI, and For You
The Framework
The PEP Model
Fiduciary duties don't operationalize in the abstract. The PEP Model maps each duty to a role, and each role to concrete embedded governance — the minimum required at every level.
PROTECT
Guardian · Duty of Care
The foundation: harm-prevention baselines. Structural safeguards that keep systems from acting against user interests.
See this duty in action — Marcus's story
ENHANCE
Mediator · Duty of Fidelity
The middle tier most frameworks skip: governance that aligns system behavior with the user's own goals — no conflicts, no self-dealing.
This is the Fidelity Gap — the middle tier most frameworks skip.
See this duty in action — Margaret's story
PROMOTE
Advocate · Duty of Loyalty
The highest tier: an affirmative best-interest obligation. Systems that actively advance user interests — including ones the user hasn't yet articulated.
See this duty in action — Eleanor's story
The Trust Feedback Loop
Fiduciary governance isn't only ethically superior — it's commercially advantageous. Each turn of the loop strengthens the next.
The loop
each turn strengthens the next.
A company adopts fiduciary governance.
Its systems owe users genuine loyalty.
Users' trust deepens.
Users engage more — and choose to pay more.
Revenue per user rises.
Company reinvests in governance.
01 · START
A company adopts fiduciary governance
02
Its systems owe users genuine loyalty
03
Users' trust deepens
04
Users engage more — and choose to pay more
05
Revenue per user rises
06
Company reinvests in governance
and the loop begins again
Asked first, most people say no.
When Apple made tracking a genuine choice rather than a default, the share of US users who could be tracked fell from 73% to 17%. Where tracking stayed the default, it stayed near 94%.
Kraft, Skiera & Koschella, Economic Impact of Opt-in versus Opt-out Requirements for Personal Data Usage: The Case of Apple's ATT — working paper, 2023, measured 2021–22.
People already pay for privacy.
$5–$10 a month is what people pay today for services that simply decline to monetize them — with no duty owed and no recourse. That is the floor, not the ceiling. Meanwhile, the ad model earned about $233 a year per person from advertising built on behavioral data — at no explicit price, with no duty attached.
Published pricing: Kagi $5–$25/mo · DuckDuckGo $9.99–$19.99/mo, checked July 2026 · Meta Q4 2024 earnings report / FY2024 Form 10-K (US & Canada; the last year Meta disclosed this figure by region).
It's already law.
In May 2026, Utah enacted the first US statute imposing a duty of loyalty on digital-identity intermediaries — binding wallet providers and verifiers not to process a person's identity attributes against their interests.
Utah SB 275, effective 6 May 2026. Analysis: Utah's SB 275 Turns Digital Identity Principles Into Law · Utah Passes Nation's Strongest Digital Identity Bill.
The GliaNet Fiduciary™ Pledge
and Net Fiduciary™ Certification Pathway
The formal instrument through which companies voluntarily adopt enforceable duties of care, fidelity, and loyalty toward their users — and the tiered certification pathway that verifies it.
The Pledge
The GliaNet Fiduciary™ Pledge is a voluntary, legally significant commitment. When a company takes the Pledge, it incorporates fiduciary duties into its own terms of service — creating enforceable, contractual obligations toward its users. Not a marketing statement. Structural accountability.
How it works
- A company or organization voluntarily adopts the Pledge, tailoring it to their business model.
- It incorporates it into user-facing terms of service and other public governance documents.
- The GliaNet Fiduciary™ pledge mark is added to the entity's website, signaling a first step commitment towards independent certification through the Net Fiduciary™ Certification Pathway.
The Net Fiduciary™ Certification Pathway
Certification progresses through five tiers, each mapped to the PEP Model. Each tier generates measurable value and builds on the commitments of the tier before it.
- The Entry Point
Tier 1
GliaNet Fiduciary™
VOLUNTARY PLEDGE
The company formally adopts the Pledge and begins embedding fiduciary duties in its governance structure.
Tier 2
Caretaker
CARE / PROTECT
Harm-prevention baselines in practice: structural safeguards implemented, operating, and measurable.
Tier 3
Partner
FIDELITY / ENHANCE
The Fidelity Gap closes: governance that aligns system behavior with the user's own goals — no conflicts, no self-dealing.
Tier 4
Advocate
LOYALTY / PROMOTE
The affirmative best-interest obligation: systems that actively advance user interests, demonstrably.
Tier 5
Steward
FULL PEP · CERTIFIED NET FIDUCIARY™ · NFSB
Full Net Fiduciary™ status — and a seat in governing the certification system itself, through the Net Fiduciary™ Standards Board.
The GliaNet Fiduciary Pledge (Tier 1) is available to all members of the GliaNet Alliance. Joining the Alliance is free of charge.
Independent certification oversight will be provided by the Net Fiduciary™ Standards Board (planned).
* The Net Fiduciary™ Standards Board and its audit regime are in development.
In Action
Beyond principles.Into execution.
Since founding, the Alliance has filed formal responses to government consultations in the United States and United Kingdom, engaged at more than sixteen institutional venues across three continents, and entered its first execution year.
Policy record
- Formal response to the White House OSTP AI Action Plan (March 2025)
- UK DSIT Call for Evidence on Data Intermediaries (May 2025) — endorsed by seventeen member organizations
Third-party recognition
- Awarded Best Collaborator at the UK Ctrl-Shift Smart Data Forum (June 2025)
Institutional reach
- 16+ engagements across three continents
- Policy, standards, academic, and industry venues — RightsCon, IEEE, Villanova, London Tech Week
Academic convenings
- “Fiduciary Duties and AI: Legal Frameworks” — NYU School of Law (June 2026)
- Health Tech & Accessibility Roundtable — Coleman Institute, CU Anschutz (January 2026)
Engage
Where do you fit?
For Companies
A public trust signal and a structured path from voluntary commitment to full embedded fiduciary governance in your business.
For Practitioners
A community of practice for legal, technical, and policy professionals researching, building, and amplifying embedded fiduciary governance into their work every day.
For Policymakers
A governance model, model bills, jurisdictional analyses, and consultation ready to serve you.
For Funders
Do you fund trustworthy digital infrastructure? We complement the work you already support.
Contact
This work moves through conversation. Funder, policymaker, company, or practitioner — if something above spoke to you, let's talk.