What We Are Finding

---

When AI companies come to Canberra: whose voice is in the room?

March 2026

Policy observation: Anthropic Australian visit and National AI Plan, March 2026

In late March 2026, one of the world’s most powerful AI companies arrived in Canberra to meet with Australia’s senior economic leadership. Infrastructure deals, investment frameworks, and national AI strategy were on the table. These are exactly the right conversations to be having. But while frontier AI was being discussed at the highest levels of government, the organisations serving Australia’s most vulnerable communities were not in the room, were not consulted, and are not resourced to keep pace with what is being decided on their behalf. Not For Humans examines what the Anthropic visit to Canberra reveals about who shapes Australia’s AI future, and what the community sector needs to claim its place in it.

ABC News, Albanese, AI and data centres, 29 March 2026.

What Happened

---

What happened and why it matters

Anthropic, the AI company behind the Claude family of models, visited Australia in late March 2026 with a delegation spanning customer, partnerships, research, and policy teams. The visit included a confirmed meeting with Treasurer Jim Chalmers and discussions with the Assistant Minister for Science, Technology and the Digital Economy. A meeting with Prime Minister Albanese was also reported. The company opened a Sydney office, marking a formal commitment to the Australian market.

This is not unusual and it is not inherently problematic. Frontier AI companies engaging seriously with national governments is exactly what the post-ChatGPT moment requires. Anthropic in particular positions itself as a safety-focused company, and its willingness to engage with government on governance frameworks rather than simply on sales is genuinely notable.

But engagement between governments and AI companies, however well-intentioned, is a specific kind of conversation. It centres on infrastructure, investment, national competitiveness, and the architecture of regulation. It asks how Australia can capture the economic opportunity of AI while managing risk at a national scale.

The Absent Voice

---

The conversation that is not happening

What is not on the agenda when an AI company meets with a Treasurer is the organisation running a food bank on a part-time coordinator and a volunteer roster. Or the community legal centre fielding calls from people who cannot navigate an automated government system. Or the mental health service trying to understand what its case management software is now doing with client data since the last product update. Or the family violence service whose intake process has just been quietly reshaped by an AI feature nobody switched on deliberately.

These are not edge cases. They are the everyday operating reality of organisations sitting closest to communities experiencing:

• The sharpest end of economic disruption and labour market displacement
• Housing stress, health inequality, and entrenched social exclusion
• Increasing demand for services at exactly the moment AI is reshaping how those services are delivered
• The greatest exposure to algorithmic decision-making in government systems they depend on

The organisations serving those communities are not resourced to participate in national AI policy conversations. They do not have policy teams. They do not have technology leads. They are often not even aware that tools already embedded in their systems have crossed a threshold that will carry compliance obligations from December 2026.

The policy context

The Missed Opportunity

---

Being left behind is not the same as being protected from harm

Here is what the “keep Australians safe” framing in the National AI Plan does not fully capture. The risk for community organisations is not only that AI will be used badly. It is that AI will be used well, by better-resourced organisations, while community organisations fall further behind.

AI genuinely offers the for-purpose sector something it has always needed. The ability to do more with less. Consider what is now within reach:

• Reducing administrative burden on overstretched staff so more time goes to direct service
• Improving how client information is gathered, organised, and acted on across complex case loads
• Extending the reach of services without proportionally increasing cost
• Supporting better funding submissions, reporting, and evidence of impact
• Enabling smaller organisations to operate with systems previously only available to large ones

These are not trivial benefits for organisations operating at the margins of financial viability. But realising them requires capacity that most community organisations do not currently have and that the market will not build for them unprompted.

The Advocacy Agenda

---

This is what Not For Humans is calling for. Not as a wish list, but as a specific and achievable set of investments that would meaningfully change the community sector’s position in the AI transition:

What You Can Do

---

What the community sector can do right now

The Anthropic visit is a useful signal, not because it threatens community organisations directly, but because it marks the pace at which the policy environment is moving. AI governance frameworks are being shaped now. The window for community sector voice is not closed, but it is not standing open indefinitely. Using that voice effectively requires:

• Peak bodies and advocacy organisations with enough AI literacy to represent the sector credibly in policy conversations
• A shared language across the sector for what AI governance actually means at the organisational level
• Organisations that understand not just that AI is arriving, but what it offers, what risks it carries, and what governance looks like at their scale
• Community organisations building their own evidence, documenting what is happening in their systems and with their workforces, so that evidence exists when it is needed

Key references