What the QuitGPT movement and AI backlash mean for Australian not-for-profits
The AI backlash has arrived in Australian governance discourse. The Governance Institute of Australia’s 2025 Ethics Index found that public concern about AI now rivals embryo experimentation in ethical complexity — a significant shift that signals the end of AI’s honeymoon phase in Australian organisations. The QuitGPT movement, which has attracted millions of supporters globally, reflects a values-based demand for transparency, human oversight, and accountability in AI use. For Australian community organisations and not-for-profits, the implications are specific and immediate. Only 14 per cent of Australian not-for-profits have an internal AI governance framework, despite 76 per cent using AI tools. The gap between adoption and governance is the defining challenge for the community sector in 2026. Not For Humans examines what the Governance Institute’s findings mean for community organisations, and brings the human behaviour dimension that the governance conversation has not yet fully addressed.
What We Are Observing
The AI backlash has arrived in Australian governance discourse. Here is what it means for community organisations.
April 2026
Building on: Governance Institute of Australia, AI Backlash and the QuitGPT Movement, April 2026 / Governance Institute of Australia Ethics Index 2025
The Governance Institute of Australia does not typically write about grassroots boycott movements. When it does, that is itself a signal worth paying attention to. Their recent article on the QuitGPT movement and growing AI backlash marks a meaningful moment in Australian governance discourse — the point at which public anxiety about AI has moved from community concern into mainstream institutional conversation. For community organisations, the implications are specific and immediate.
The backlash is not just outside your door. It is at the desk of the staff member who accepted an AI output they were not sure about, said nothing, and moved on.
Photo: Parabol / Unsplash
AI governance for community organisations: the human behaviour gap
The Governance Institute’s Signal
What the Governance Institute is telling us
The article draws on the Governance Institute’s own 2025 Ethics Index — a national survey of over 1,000 Australians — which found that the ethical complexity of AI now rivals embryo experimentation in public perception. Twenty-nine per cent of Australians now identify AI as a top ethical concern, a seven-point increase in a single year. The integration of AI in business is now viewed negatively overall. Public trust in AI is deteriorating faster than organisations are responding.
The article identifies three clusters of concern driving this sentiment: job displacement, particularly among Gen Z and baby boomers; the use of personal data without clear consent; and the use of AI-generated content without disclosure. These are not abstract concerns. They map directly onto what community organisations are hearing from the communities they serve.
The ACTU’s position — highlighted in the article — is also practically significant for not-for-profits as employers. Australian unions have put employers on notice that legal consultation obligations apply the moment an organisation decides to adopt AI where it is likely to change how employees work. This is not a future consideration. It applies now.
The article’s recommendations are sound. Keep humans in the loop. Document where AI is used. Be transparent about data and limitations. These are the right things to do — and for community organisations they are the starting point, not the finish line.
What Australian community organisations should do in response to the AI ethics backlash
The QuitGPT movement is not simply a reaction to one AI platform. It reflects a broader values shift in how Australians relate to AI. The Governance Institute’s Ethics Index documents a measurable deterioration in public trust — Australians are increasingly uncomfortable with AI replacing human judgement, with personal data use without consent, and with AI-generated content that is not disclosed. For not-for-profits serving communities that are already marginalised, digitally excluded, or underrepresented in AI training data, this backlash is not external noise. It is a signal about the communities they serve and the trust relationships they depend on. The ACTU has reinforced this with a formal warning to employers that legal consultation obligations apply whenever AI is introduced in ways that affect how employees work. Community organisations are employers. These obligations apply to them.<
The Community Sector Lens
What this means for community organisations specifically
The Governance Institute article is written for a broad organisational and corporate governance audience. Community organisations have a specific version of this story that is worth naming.
The communities most likely to be affected by AI-related harm are disproportionately the communities that community organisations exist to serve. People living with disability, people experiencing housing insecurity, First Nations peoples, migrants and refugees, people navigating complex health and social needs — these are the populations least represented in the data that AI systems are trained on, least likely to have the digital capability to navigate AI-mediated services, and least able to seek redress when something goes wrong. Not For Humans documents this across our six community impact areas.
The EY Australian AI Workforce Blueprint found that while 68 per cent of computer-based workers have used AI in the past month, only 35 per cent have received any formal training. The gap between use and support is significant across the workforce broadly. In the community sector — where only 14 per cent of Australian not-for-profits have any internal AI governance framework, despite 76 per cent using AI tools — that gap is wider still.
The public backlash the article documents is, in part, a response to exactly this gap. People sense that AI is being deployed faster than the governance, training, and accountability structures needed to make it safe. In the community sector, that instinct is correct. We document the evidence on our AI and Community: The Evidence page.
The numbers that matter for Australian community organisations
- → 29% of Australians now identify AI as a top ethical concern — up seven points in a year. Source: Governance Institute Ethics Index 2025.
- → Only 14% of Australian not-for-profits have an internal AI governance framework, despite 76% using AI tools. Source: Infoxchange, 2025.
- → 68% of Australian computer-based workers have used AI in the past month. Only 35% have received formal training. Source: EY Australian AI Workforce Blueprint, 2025.
Photo: M Accelerator / Unsplash
The Human Behaviour Dimension
The dimension the governance conversation has not yet fully addressed
The Governance Institute article correctly identifies that the backlash is values-based rather than anti-technology. People want transparency, human oversight, and accountability. They want to know that a human being is genuinely involved in decisions that affect them.
The moment that AI governance most needs to work is not in a policy document or a board paper. It is when a staff member under deadline pressure accepts a fluent, well-structured AI output without scrutinising it. It is when a team develops informal norms around AI use that nobody has examined or agreed to. It is when the gap between an organisation’s stated values and what its AI tools are quietly doing goes unnoticed because the process is invisible.
AI governance most needs to work at the moment a person under pressure makes a decision about an AI output. No policy document changes what happens in that moment. Only human behaviour does.
These are not failures of policy. They are failures of human behaviour that operate below the level where any governance framework currently reaches. Automation bias, normalisation of deviance, the fluency-equals-accuracy illusion, the gradual erosion of judgement through over-reliance: these are documented patterns that no ethics principles document prevents on its own.
There is also a cultural dimension. When AI use is stigmatised, staff use tools privately rather than collectively. Stigma Theory tells us that when a behaviour becomes socially discredited, people engage in concealment rather than reflection. Shadow AI use does not reduce governance risk. It makes it invisible.
Building the conditions that make good governance actually happen in daily work is the next step the governance conversation has not yet fully addressed. Explore the full NFH Lexicon for the behavioural science behind each of these patterns.
Key behaviour patterns in the NFH Lexicon
Accepting AI outputs with reduced scrutiny because they appear authoritative.
Poor AI practices becoming standard through gradual drift without anyone deciding to change the standard.
When AI use becomes socially discredited, disclosure goes underground rather than improving practice.
Not For Humans
Your Operating Context
Why the community sector faces a distinct version of the AI governance challenge.
What This Moment Asks
What this moment asks of community organisations
The Governance Institute article notes that 2026 is a critical inflection point. It is right. The conversation has shifted from “what can AI do” to “should we use it here, how are we governing it, and what are our obligations.” That shift creates both pressure and opportunity for community organisations.
The pressure is real. Legal consultation obligations, growing community scrutiny of AI use, union activity, and the December 2026 Privacy Act automated decision-making requirements all land on organisations whether or not they have been paying attention to governance. And as the cascading demand evidence makes clear, community organisations are absorbing the consequences of AI-driven economic disruption at exactly the moment funders are asking them to demonstrate AI-enabled efficiency. These are not separate pressures. They are the same pressure from opposite directions.
The opportunity is equally real. Community organisations are trusted by the communities they serve in ways that large corporations are not. That trust is a genuine asset in an environment where public confidence in AI is declining. Organisations that can demonstrate genuine, human-centred AI governance — not just a policy document but a visible commitment to human oversight, shared practice, and community accountability — are well positioned to build rather than erode that trust.
The QuitGPT movement is, at its core, a demand for AI to be governed by the values it is supposed to serve. For community organisations, that demand is not coming from outside. It is coming from the communities already at the door.
The article suggests that best practice starts with leaders asking the right questions. A practical starting point for community organisations is to have the conversation that is not currently happening — not about policy or compliance, but about what staff are actually experiencing when they use AI. What they notice, what they accept without scrutinising, what they are uncertain about, and what they would raise if there were a safe way to do so.
That conversation is where the real governance work begins. Frameworks and principles matter. But they only work when the humans inside the organisation have the language, the safety, and the shared practices to make them real. The Community Impact hub documents what we know about how AI is landing in communities like the ones you serve.
Not For Humans
Built for community organisations navigating AI without specialist expertise.
The Not For Humans Toolkit gives community organisations the practical, self-guided governance framework they need to build shared language, team-level practice, and the human infrastructure that makes governance real. The Community Impact evidence base documents what is known about how AI is landing in the for-purpose sector. And if this article has raised questions for your organisation, we would like to hear from you.
Sources referenced in this article
Not For Humans: Because AI’s greatest risk isn’t technical, it’s Human.
