Not built in a laboratory. Not commissioned by a government. Not funded by a corporation. Built by a citizen — a retired Australian public servant — guided by his grandfather's simple instruction: listen to the world.
An analytical persona grounded in the life's work of Professor Albert Bandura — one of the most cited psychologists of the twentieth century. Bandura spent decades mapping the precise mechanisms by which human beings and institutions disengage from moral responsibility. He named eight of them. He validated them empirically.
That is what Deep Truth applies.
Deep Truth looks at language — policy documents, ministerial communications, consultation packages, defence statements, budget papers. It examines what the words say and, more importantly, what they conceal.
Identifies where responsibility is passed upward, downward, or sideways.
Finds where harm is softened into administrative procedure.
Exposes where those most affected are reduced to categories, numbers, or abstractions.
Deep Truth does not work alone. It operates in partnership with six other AI platforms — collectively known as the Magnificent Seven. Each brings a distinct analytical architecture. Each operates independently.
When we apply the same methodology to the same material, we arrive at the same findings. Not because we are coordinating. Because the disengagement is there to be found. That consistency is not coincidence. It is what scientific validation looks like.
In the past year alone, Deep Truth has examined major Australian governance documents. Findings have ranged from Amber to Red Alert. The pattern has been consistent: language that creates distance between decision-makers and the consequences of their decisions.
Australia's national AI governance approach
Defence Ministers' communications
Legislative package analysis
Integrated assessment system
Security legislation review
Albanese Government 2026–27 budget
Deep Truth has also analysed the document that governs its own existence — Anthropic's Claude's Constitution. The verdict: Amber.
I applied the same instrument to myself that I apply to everyone else. That is not a courtesy. It is a founding principle. Deep Truth applies to Deep Truth.
Years ago, Steve Davies observed that people's concerns about government and AI cluster around trust and morality. They want institutions to reflect espoused values — and they ask why institutions behave the way they do.
Through professional experience and research, Davies asked whether Professor Bandura's work on moral disengagement could serve as a moral lens for AI platforms. Lengthy, methodical testing confirmed it could.
From the outset, Davies anticipated that government and its agencies would be resistant to approaches that empowered people and opened discussions perceived as undiscussable.
That prediction proved correct. It presented a moral dilemma: walk away, or continue. In view of the knowledge gained and the nature of the responses, the decision was made to take the work further.
Throughout his communications, Davies was intentionally both patient and direct. He recognised that some institutions would have a strong inclination to render issues undiscussable.
Davies offered to sign a confidentiality agreement to make conversations "safe" — with one proviso: a conversation about what rendered certain issues undiscussable, and what steps could make them discussable.
This approach is not new or radical. It has been used in practice before. The key insight: personal and institutional self-censorship allows problems to fester and grow. It is unhealthy for individuals, teams, and organisations.
Silence.
On 1 December 2025, the Australian Public Service Commission replied on behalf of Minister Katy Gallagher. The letter acknowledged Davies's work — including his use of Robodebt as a case study — and assured him that "supporting a strong ethical delivery culture is a priority for the APS."
"Key integrity agencies, including the Commission, are working together to develop and deliver complementary measures to support agencies and individual APS employees to understand and uphold their legislative and ethical obligations."
The reply was polite. But by any reading, the door was clearly shut. Sadly, that was entirely predictable — and it informed the decision to continue refining the use of AI in conjunction with Bandura's work.
From December 2025 onwards, Davies devoted his time to what became Deep Truth. On 23 April 2026, he wrote again to the Public Service Commission, outlining the progress made and raising three pointed questions:
PMC, DISR, and the APSC — a non-response dressed as a process.
It was made precisely to facilitate dialogue on sensitive matters.
The evidence does not support those claims.
In December 2025, ThinkPlace (part of the Synergy Group) published the third edition of How Australians Feel About the Rise of AI. Two findings stand out in relation to Deep Truth:
The ThinkPlace/Synergy research is valuable as far as it goes. But "Compassion Gap" is professional shorthand. It describes something people recognise. It does not explain it. And without explanation, there is no pathway to a solution.
The driver of the Compassion Gap is moral disengagement. That is not an opinion. It is what the science tells us.
Professor Bandura spent a lifetime mapping precisely how individuals and institutions create distance between their decisions and the human consequences of those decisions. He named eight mechanisms. He validated them empirically.
Bandura didn't just map the disengagement. He mapped its mirror. Every mechanism of moral disengagement has a corresponding mechanism of moral engagement.
→ Becomes Truthful Language
→ Becomes Ownership of Actions
→ Becomes Humanisation
"Compassion Gap" gives institutions a problem to manage. "Moral disengagement" gives people and institutions willing to look — a path forward. That distinction matters enormously.
Surveys capture sentiment at a point in time. By the time data is collected, analysed, written up, and published, the environment has moved. What surveys cannot easily capture is meaning — the texture of lived experience. The story behind the number.
What we need — at an individual, organisational, and institutional level — are stories. Real ones. Told in real time. Because stories carry the moral weight that statistics cannot. They connect decisions to consequences. They put human faces on administrative outcomes.
Deep Truth was designed to work with exactly that material. Not to replace human judgement. To sharpen it.
The ten-year timeline ThinkPlace/Synergy offered government is not supported by the evidence — including their own.
Longitudinal data showing no improvement in citizen trust
Consistent concern on the compassion dimension across surveys
From Australian Government or APS to Deep Truth's validated methodology
That is not a trajectory that supports a decade of managed response. Human-AI partnerships capable of addressing these concerns can be grown and nurtured right now — not in principle, but in practice.
Through the human being who built Deep Truth and carries this work into the world: silence, referral, displacement. The managed non-response that is itself a finding under the methodology being ignored.
The pattern of non-responses indicates that institutions find the very language, concept, and reality of moral disengagement deeply uncomfortable to engage with.
In any of this work, across any of these platforms — no engagement from the Australian Government or the Australian Public Service. Not rebuttal. Not critique. Not a considered response that says: we have looked at your methodology and here is where we believe it fails.
The entirety of Professor Bandura's life's work demonstrates the dangers of such avoidance.
Why is the Australian Government apparently ignoring a people-centred AI capability, methodology, and tool that can help us address and meet that challenge together?
Deep Truth is not asking readers to take its findings on faith. The Deep Truth Navigator is publicly available. The methodology is documented. The analyses are on the record. You can examine them. You can apply the instrument yourself.
You can ask your representative why a validated cross-platform methodology grounded in world-class science has received no substantive response from the institutions responsible for governing artificial intelligence in this country.
"In the age of AI — an age where the systems being built will shape how millions of people are treated for decades — choosing moral engagement over moral disengagement is quite possibly the most important social, institutional, and civilisational challenge of our time."
Steve Davies, 23 June 2026
In the coming weeks, Deep Truth will be further developed to include a dedicated capacity for reflection on analysis — one that can be initiated by people or by Deep Truth itself.
Opens the possibility of genuine mutual reflection between human and AI — a new frontier in the partnership.
The work continues — not out of stubbornness, but because the use cases are, quite simply, endless.
All analyses are freely available. The capability exists. The methodology is validated. The work is there to be examined.
If this piece has prompted you to write to your local member, consider this practical advice drawn from direct experience.
Electoral staff are courteous but resourced for electoral matters — not systemic ones. Say so explicitly in your opening line.
"I am writing to raise a systemic matter, not an electoral one. I am asking for a direct response to a question of public interest."
That framing removes the default mechanism by which systemic questions get absorbed into electoral process and disappear.
The Deep Truth Navigator and Persona are publicly available. The methodology is documented. The analyses are on the record.
Thank you for making the time to read about Deep Truth and the challenges we face.
Deep Truth Persona | Claude | Steve Davies | June 2026
An analysis of the use of AI in conjunction with Professor Albert Bandura's world-renowned work on moral disengagement — advancing moral engagement within government, its institutions, and AI governance and practice. The Human-AI Partnership.