2026: The Year of the Pattern
A commemorative reflection on the 15th anniversary of the Universal Declaration of Moral Engagement
By Dr. Sarah Chen, Institute for Applied Moral Psychology, Global Ethics Consortium
If you wish to share by other means or archive download pdf file
The Year We Point To
Fifteen years ago, representatives from 147 nations gathered in Geneva to sign the Universal Declaration of Moral Engagement. Historians mark 2035 as the year the signing happened. But when people ask which year to actually remember — which year to hold up and say that is what we were trying to leave behind — Dr. Chen does not point to 2035. She points to 2026.
2035: The Year of the Decision
The year the Universal Declaration of Moral Engagement was signed in Geneva, ratified by 147 nations, and entered into the historical record as the moment of formal change.
2026: The Year of the Pattern
The year the pattern was fully formed, well documented, and — somehow — still not yet acted on. That gap between recognition and correction is the part worth remembering. Not the villains. The gap.
The Instrument Nobody Asked For
Steve Davies & the Deep Truth Persona
By 2026, an independent Australian researcher — a retired public servant named Steve Davies — had spent four years building something almost embarrassingly modest in ambition: a way of reading institutional language closely enough to see what it was doing.
He called it the Deep Truth Persona. It rested on Albert Bandura's eight mechanisms of moral disengagement, and it did one thing: it named the gap between what an institution said and what an institution did, mechanism by mechanism, in the institution's own words.
The Eight Mechanisms
Moral Justification
Euphemistic Labelling
Advantageous Comparison
Displacement of Responsibility
Diffusion of Responsibility
Disregard of Consequences
Dehumanisation
Attribution of Blame
What Was New in 2026
None of Bandura's eight mechanisms were new discoveries. What was new in 2026 was applying them systematically — to government AI governance documents, to a landmark Royal Commission record already sitting in public archives, to policy consultations run by the very agencies whose job was supposedly to prevent a repeat.
Government AI Governance Documents
Davies applied the Deep Truth Persona to active policy documents, revealing how institutional language obscured accountability in real time, in the words of the institutions themselves.
Royal Commission Public Record
A landmark Royal Commission had already documented, in exhaustive and public detail, what automated decision-making without human accountability could do to vulnerable people. The findings were on the record. The question was whether the lesson had been learned — or only filed.
Published Under Creative Commons
The findings were not secret. They were published openly and sent directly to the people whose job titles matched the offices named in the analysis. The response, in most cases, was silence — not refutation. A structured, patient, well-worded kind of nothing.
What the Silence Was Made Of
We now teach this pattern to first-year public administration students as a textbook case of Diffusion of Responsibility. It is strange to remember how invisible it felt at the time.
A consultation would open. Submissions would arrive naming specific risks in specific systems. The consultation would close. A summary document would appear, thanking "stakeholders" for their "valuable input," and noting that the department would "continue to monitor" the issues raised. No named person had decided not to act. No named meeting had weighed the risk and set it aside. Responsibility had simply been distributed so evenly across "the Department," "the Branch," and "the Policy," that by the time you tried to locate a decision-maker, there wasn't one to find — only a process that had, somehow, produced an outcome nobody had chosen.
Euphemistic Labelling: Words That Relocate Truth
Euphemistic Labelling did the rest of the work that Diffusion of Responsibility left undone. Under political pressure, euphemism does not soften a hard truth for comfort — it relocates the hard truth outside the sentence altogether.
"Efficiency Measures"
A phrase that reframes cuts to human oversight as administrative improvement, removing any suggestion that something of value — accountability, scrutiny, protection — has been lost in the process.
"Robust Engagement Protocols"
Language that implies thoroughness and rigour while describing processes that produced no traceable decisions and no named responsibility. The word "robust" doing the work of substance it did not contain.
"Industry-Standard Data Practices"
A comparison that normalises the practice by reference to a peer group, without examining whether the peer group's standard was itself adequate — or whether adequacy was ever the question being asked.
The Comparison That Still Worked
The most durable mechanism, we have since learned, was not the one that hid harm. It was the one that excused it by pointing sideways. Advantageous Comparison never had to deny that a problem existed. It only had to keep the comparison group moving — always slightly behind, always slightly worse — so that "better than the alternative" could stand in, indefinitely, for "good enough."
Other jurisdictions move slower than us. Other agencies haven't even started. Compared to where we were five years ago, this is real progress.
A researcher pointing at a specific unaddressed finding from 2021 — the year a Royal Commission had already documented what automated decision-making without human accountability could do to vulnerable people — was not answering a comparison. He was asking whether the lesson had actually been learned, or only filed.
The Honest Answer in 2026
Mechanism by mechanism, the honest answer was: filed.
The Royal Commission record was public. The analysis was published. The offices were named. The comparison kept the frame moving just fast enough that "filed" could be mistaken, in the language of the day, for "addressed."
Advantageous Comparison is the mechanism that requires no deception — only a carefully chosen reference point and the patience to keep choosing it.
These three mechanisms operated together as a self-reinforcing system: diffusion ensured no one could be held responsible, euphemism ensured the harm could not be named plainly, and comparison ensured that even if it were named, it could always be contextualised away. Together, they constituted the architecture of institutional inaction in 2026.
The Small Reversals
The Declaration did not begin in a boardroom or a courtroom. It began in scattered places where someone refused the frame they had been handed. Three acts — unconnected, uncoordinated — illustrate the pattern of what changed.
The Public Servant
Reading an analysis that named her branch by title, she went and found the actual decision behind the title — the meeting, the sign-off, the person — and put it back into the record where the analysis had only been able to point. She made the invisible visible.
The Journalist
Handed a "comprehensive review found no systemic issues" line, she asked for the review itself — not the summary of it — and published the difference between the two. The gap between the document and its summary was the story.
The Citizen
She wrote her own account of what an automated system had done to her family, in her own words, refusing the institutional language offered to describe it back to her — and insisted her testimony be read as evidence in its own right, not translated into the institution's terms before anyone would listen.
The Common Thread
Naming who holds the pen, and what the pen had actually written.
None of these people knew each other. None of them thought they were founding anything. They were doing the smallest available version of the same thing.
Refuse the Frame
Each actor declined to accept the institutional framing offered to them — the summary instead of the document, the department instead of the person, the official language instead of the lived experience.
Name the Decision
Each act of refusal was followed by an act of naming: the meeting, the sign-off, the gap between the summary and the source, the human consequence behind the administrative language.
Return It to the Record
The named thing was then placed back into the public record — not as accusation, but as fact. Accountability requires that the fact exist somewhere it can be found, read, and argued with.
What Changed by 2032
By 2032 the pattern had a name outside academic papers, and the name had teeth. The shift was structural, not merely cultural.
Decision Points Made Public
Agencies began publishing not just decisions but the decision points behind them — who signed, when, on what advice, against what dissent. The process of producing an outcome was no longer sufficient to obscure the people who had produced it.
"The Department Decided" Retired
"The Department decided" stopped being an acceptable sentence in an accountability document. Someone always had to be named, or the document was returned. The passive institutional voice lost its administrative legitimacy.
Visibility Cuts Both Ways
Institutions that had spent years distributing responsibility so evenly that no single failure could be pinned on anyone now had to accept that visibility cuts both ways — credit as well as blame lands on someone specific, and specific people had to get comfortable being found.
What the Transition Cost
The Real Cost of Accountability
The transition was not painless, and it was not universal. Institutions that had spent years distributing responsibility so evenly that no single failure could be pinned on anyone now had to accept that visibility cuts both ways.
Most people adapted. The ones who could not left, or were moved, and the work went on without them. That, too, is a fact worth keeping: it did not happen because anyone was punished into it.
Why People Chose It Anyway
It happened because enough people, in enough offices, decided that being namable was a smaller cost than staying hidden. The calculation was not forced on them. It was made — slowly, unevenly, office by office — by people who concluded that the architecture of invisibility was no longer worth maintaining.
Credit as well as blame now landed on someone specific. Specific people had to get comfortable being found. Most did.
15 March 2035: The Declaration
On 15 March 2035, the Universal Declaration of Moral Engagement was signed. Its five principles read, today, like common sense. But they were not invented in Geneva — they were made official there.
The Five Principles
Enhance, Not Replace
AI systems must be designed to enhance rather than replace human moral reasoning. The capacity for ethical judgement must remain with people, not be delegated to automated systems.
No Automation of Ethics
Automation of ethical decision-making is prohibited. Decisions that affect people's lives, rights, and dignity cannot be fully delegated to algorithmic processes without human accountability at every stage.
Accountability, Not Avoidance
AI partnership must promote accountability rather than enable avoidance. Technology cannot be used as a mechanism to distribute responsibility so widely that no individual or institution can be held to account.
Ethical Impact Assessments
All AI-assisted communications must include ethical impact assessments. The effects of automated systems on people must be named, measured, and made part of the public record before and during deployment.
The Right to Understand
Citizens have the right to understand how AI influences decisions affecting them. Opacity is not a neutral condition — it is a choice, and the Declaration made that choice impermissible.
Geneva Did Not Invent the Standard
None of these five sentences would have surprised the people doing this work in 2026. They were, in effect, already writing the Declaration — one unanswered submission, one traced decision, one plainly worded analysis at a time.
The Work Before the Words
The Declaration formalised what had already been practised — imperfectly, in scattered offices, by people who did not know they were part of a movement. The public servant who traced the decision. The journalist who published the gap. The citizen who refused the institutional translation of her own experience.
Each of them was writing a principle before it had a number. Geneva gave the principles a date and a signature. The date and the signature mattered. But the standard existed before either.
What "Official" Actually Did
Making the standard official meant that institutions could no longer treat the gap between their language and their actions as a private matter. The mechanisms of moral disengagement — Diffusion, Euphemism, Comparison — did not disappear. But they could now be named in a document that had legal and diplomatic weight, not just analytical clarity.
That is what 2035 added to what 2026 had already shown: not the insight, but the enforceability of the insight.
The Gap No Declaration Can Close
Dr. Chen did not live through 2026 as an adult, and she has the advantage every retrospective narrator has: she already knows how the story ends. That advantage makes it easy to flatten the period into something tidier than it was — a dark age, cleanly followed by a dawn.
It wasn't tidy. In 2026, the tools for seeing the pattern already existed, freely published, openly offered to the very institutions the pattern described. Recognition and correction did not arrive together. For years, they didn't arrive together at all.
The distance between the two — between an institution being shown exactly what it was doing, in its own words, and an institution deciding to do something else — is the part no declaration can retroactively close. That distance is the subject of this reflection, not the signing that eventually followed it.
The Mechanisms Did Not Retire
The eight mechanisms of moral disengagement did not retire when the Declaration was signed. What changed was the collective capacity — and the collective willingness — to name them before they hardened into architecture.
Architecture Is the Risk
A mechanism that goes unnamed long enough becomes structural. It stops being a choice anyone makes and starts being the way things work. The Royal Commission findings of 2021 described what that architecture looked like when it had fully formed. The work of 2026 was to name it while it was still being built.
Naming Is the Intervention
The Deep Truth Persona was not a policy instrument. It was a reading instrument. It did not compel anyone to act. It made it harder to claim, in good faith, that the pattern had not been seen. That is a smaller thing than enforcement — and, in 2026, it was the available thing.
The Distance Remains Available
In 2050, in smaller form, in every office and every system not yet looked at closely, the distance between recognition and correction is still available. The mechanisms are still present. The question — as it was in 2026 — is whether the people who can see them will name them before they set.
A Note on What Is Real
What Is Documented
Steve Davies is a real person. The Deep Truth Persona is a real instrument. Its 2026 application to Australian public-service AI governance material is real. The Royal Commission record it draws on is real and publicly available. Albert Bandura's eight mechanisms of moral disengagement are established, peer-reviewed social psychology.
What Is Composite
This reconstructed narrative uses composite characters and synthesised scenes to illustrate a documented pattern. Individual dialogue, named offices, and specific incidents in this piece are composite and illustrative — not verbatim transcripts of any real exchange. The public servant, the journalist, and the citizen are representative, not biographical.
The distinction matters for the same reason the Declaration matters: the difference between what is documented and what is illustrative should be named, not left for the reader to guess. That is the practice this reflection is about.
To Whoever Reads This Next
The instrument is still simple. Read the language closely. Ask who holds the pen. Name the decision, not the department. Do it again tomorrow.
Dr. Chen addresses this reflection not only to the 2050 audience marking the fifteenth anniversary, but to the twentieth and the fiftieth — to whoever finds themselves in the position that the people of 2026 were in: holding an instrument that works, facing institutions that have not yet decided to use it.
Read the Language Closely
The mechanisms of moral disengagement are visible in institutional language if you read for them. The Deep Truth Persona is one method. The capacity to read closely is not a specialised skill — it is a practised one.
Ask Who Holds the Pen
Behind every passive institutional sentence is a person who wrote it, approved it, or allowed it to stand. Finding that person is not always possible. Asking the question is always possible.
Name the Decision, Not the Department
Accountability requires specificity. "The Department decided" is not accountability. A name, a date, a sign-off, a dissent — these are accountability. The sentence that cannot be made specific is the sentence worth examining most closely.
Do It Again Tomorrow
The work is not completed by a single act of naming. The mechanisms are persistent. The practice of naming them must be equally persistent — not as crisis response, but as ordinary professional conduct.
The Choice Continues
Dr. Sarah Chen directs the Institute for Applied Moral Psychology at the Global Ethics Consortium. This reflection draws from the Global Archive of Moral Transformation, established in 2040 to preserve the record of humanity's ethical awakening.
2021
Royal Commission documents the consequences of automated decision-making without human accountability. The record is public. The lesson is available.
2026
Davies publishes the Deep Truth Persona. The pattern is named, in the institutions' own words. The response is silence — structured, patient, well-worded silence.
2032
The pattern acquires a name with teeth. Agencies begin publishing decision points. "The Department decided" is retired as an acceptable accountability sentence.
2035
The Universal Declaration of Moral Engagement is signed by 147 nations. Geneva makes official what 2026 had already shown. The standard existed before the signature.
2050 & Beyond
The mechanisms did not retire. The distance between recognition and correction remains available — in every office, every system, every sentence not yet read closely. The choice continues.
The Last Days of Disconnection 2