Australia is quietly building an architecture of total surveillance, and disabled people may be among its first targets
Palantir does not need to know everything about you, it simply needs access to the systems that do.
Your bank knows where you spend money. Services Australia knows what income you receive. The NDIS knows your disability, support needs, living arrangements and daily routines. Home Affairs knows when you cross a border. Police hold intelligence about where you have been, who you associate with and whether the state considers you suspicious. Health systems know your diagnoses, medications and moments of crisis.
Palantir’s business is making those separate fragments speak to one another.
It builds the infrastructure through which governments, intelligence agencies, militaries, police and corporations can bring enormous datasets together, identify relationships, generate profiles, assign risks and direct action. Its products do not merely store information. They transform information collected for one purpose into power capable of being used for another.
In the United States, that power helps Immigration and Customs Enforcement decide whom to pursue and deport. In Israel, Palantir has supplied technology for “war-related missions”. In Australia, its software is used by Defence, AUSTRAC and Victoria’s prison system, while federal and state contracts with the company have reportedly approached $80 million.
Now there is an urgent, unanswered question: has information about NDIS participants entered Palantir’s systems through the government’s sprawling Fraud Fusion Taskforce?
The government will not say.
That is not a minor gap in administrative transparency. It is a warning that Australia is assembling an ultra-surveillance state behind redactions, procurement contracts and the deliberately soothing language of “fraud prevention”, “data analytics” and “integrity”.
A welfare system transformed into an intelligence network
The Fraud Fusion Taskforce began operating in November 2022. Led by the NDIA and Services Australia, it brings together 25 government agencies, including AUSTRAC, the Australian Taxation Office, Australian Federal Police, Australian Criminal Intelligence Commission and Department of Home Affairs.
This means information about disabled people and their support providers can circulate through a network containing taxation, financial intelligence, policing, criminal intelligence and border-enforcement agencies.
NDIS participants did not meaningfully consent to this. They were not asked whether the intimate information required to access disability support could become part of a government intelligence network. They were not given a public account of which agencies could see which information, how long it would be retained, what analytical systems would process it or what conclusions might be drawn from it.
The rules followed the surveillance.
The Taskforce’s Memorandum of Understanding was signed in May 2023, six months after the information-sharing network began operating. Its Data Governance Framework was not finalised until June 2024, eighteen months after the Taskforce commenced. Both documents have been partially released through Freedom of Information. Both are heavily redacted. The sections the public most needs to see, how information moves, where it is processed, who can access it and what technology is involved, remain concealed.
There is apparently an Ethics and Human Oversight Framework. It contains a code of conduct but does not identify the technology vendors handling the information. The framework telling government officials how to handle your data does not tell you who holds it.
This is how state surveillance expands in a liberal democracy. Rarely through one dramatic announcement declaring that the government intends to build a system capable of mapping entire populations. It arrives fragmented across agencies, contracts, trials, memoranda and taskforces. Each agency insists it sees only one small piece. Each disclosure is redacted. Each new power is framed as technical, temporary or necessary.
Then the pieces are connected. That act of connection is Palantir’s speciality.
The question AUSTRAC will not answer
AUSTRAC, Australia’s financial intelligence agency, holds a Palantir contract for data analytics reportedly worth more than $12 million.
Banks and other financial institutions are instructed to report transactions that may indicate NDIS fraud or financial crime. These reports can contain account details, transaction histories, names and other identifying information connected to NDIS participants and providers.
AUSTRAC is also part of the Fraud Fusion Taskforce. This establishes a pathway through which NDIS-related financial information reaches an agency using Palantir. It does not, by itself, prove that the information is processed within Palantir software. That is the question requiring an answer.
When Michael West Media asked AUSTRAC whether NDIS participant data reaches Palantir’s systems, AUSTRAC did not address it. The NDIA reportedly answered only one of eleven questions.
Australians are therefore left with an extraordinary situation: the government has constructed a 25-agency information-sharing network around disabled people, one participating agency uses one of the world’s most controversial surveillance platforms, and nobody in government will provide a clear answer about whether participant information enters that platform.
Secrecy does not protect the public here. It protects the surveillance architecture from public scrutiny.
Robodebt’s architects did not disappear
The echoes of Robodebt are impossible to ignore.
Two officials now holding senior positions inside the NDIA’s integrity division gave evidence to the Robodebt Royal Commission. One wrote the original 2014 concept paper that became Robodebt. The other supervised and approved it.
Robodebt unlawfully raised debts against more than 400,000 welfare recipients. It terrorised people already living on little, forcing them to disprove debts generated by a method the government knew was unlawful. People lost money, housing, health and hope. Some families told the Royal Commission that their loved ones died by suicide after being pursued.
Nobody was prosecuted. Those responsible were absorbed back into the institutions of government. Some continued working in compliance, integrity and automated decision-making, the very fields through which Robodebt was made possible.
While the Royal Commission was examining what happened, the NDIA’s integrity division was communicating with Palantir.
Freedom of Information documents record an email thread titled “NDIA/Palantir Progression”, discussions about a proof of concept and questions about Palantir’s mandatory security assessment. The company reportedly did not obtain the relevant clearance until November 2025, more than two years after an NDIA security officer queried its absence. The NDIA nevertheless withheld 22 documents relating to the engagement.
The NDIA says it has not used Palantir technology and has no plans to do so. But that does not resolve the wider question of whether NDIS-related information enters Palantir through another Taskforce agency.
Robodebt should have taught governments that welfare recipients are not datasets to be mined, scored and punished. Instead, it appears to have taught the bureaucracy to spread responsibility across more agencies, hide more information behind commercial confidentiality and install more layers between the automated system and the person whose life it destroys.
The machinery is becoming harder to see, not less dangerous.
From welfare surveillance to mass deportation
Palantir presents itself as merely a software company. It claims its customers control their own information and determine how its products are used. However, “we only build the tool” is not an ethical defence when the tool is designed to make state violence faster, broader and more efficient.
Palantir has worked with United States immigration authorities for years. In 2025, ICE awarded it a contract worth almost US$30 million to build ImmigrationOS, a platform intended to assist the selection and apprehension of people for deportation and provide “near real-time” information about people leaving the country.
This is not neutral data organisation. It is digital infrastructure for raids, detention, family separation and removal.
Amnesty International concluded in 2020 that there was a high risk Palantir was contributing to serious human rights violations against migrants and asylum seekers through its relationship with ICE. Amnesty argued that the company was failing to meet its responsibility to prevent its products from enabling abuse. Palantir nevertheless expanded the machinery.
The surveillance logic should be familiar. Information collected across different parts of a person’s life is consolidated into an actionable profile. A person becomes an address, an association, a border encounter, a visa status, a financial transaction and a “confidence score”. The system does not need to understand their life. It needs only to make them locatable.
Palantir turns the violence of deportation into a data-management problem.
Technology for Israel’s war
Palantir’s relationship with Israel destroys any pretence that its work can be separated from what its customers do.
In January 2024, after holding a board meeting in Tel Aviv, Palantir announced a strategic partnership with Israel’s Ministry of Defence. The company said the agreement would provide technology to support Israel’s war effort. A Palantir executive stated that the parties would use its technology for “war-related missions”. The company has not publicly disclosed the full nature of those operations.
That secrecy matters in the context of Israel’s assault on Gaza, where surveillance, artificial intelligence and data-driven targeting have operated alongside mass killing, forced displacement and the destruction of civilian infrastructure.
Palantir cannot simultaneously advertise that its systems deliver decisive operational advantages and then insist that it bears no responsibility when those advantages are used by militaries committing devastating violence. Its executives understand that software changes what states are capable of doing. That capability is precisely what they sell.
Palantir’s leadership has also made its politics increasingly explicit. The company, co-founded by Trump-aligned billionaire Peter Thiel, has published a manifesto extolling American power and suggesting that some cultures are “dysfunctional and regressive”. A British MP described it as the “ramblings of a supervillain”. The language is grotesque, but it is not disconnected from the product.
A company that believes in a hierarchy of cultures is selling governments the capacity to classify, monitor and act upon entire populations. A company aligned with militarism, mass deportation and Western supremacy wants access to health, welfare, policing, financial and defence systems.
We should take that ideology seriously.
The infrastructure of suspicion
Palantir’s danger is not simply that it “collects lots of data”. Governments have always gathered information about populations, particularly those they govern through punishment rather than care.
Its danger lies in integration.
A disability payment becomes financial intelligence. A hospital visit becomes a risk indicator. An immigration record becomes a target. A friendship becomes an association. An error in one database can spread across twenty others and acquire the appearance of truth through repetition.
This architecture intensifies the existing biases of the state. Aboriginal and Torres Strait Islander people, disabled people, migrants, welfare recipients, criminalised communities and people living in poverty are already subjected to disproportionate surveillance. They appear more frequently in government datasets because the state intrudes more frequently into their lives. More contact produces more records. More records generate more apparent “risk”. More risk justifies more surveillance. The system then cites the data created by over-policing as evidence that more policing is necessary.
Palantir did not invent this circular logic. It industrialises it.
The company’s platforms can make relationships visible across datasets at speeds and scales no individual analyst could achieve. That capacity is promoted as efficiency. But efficiency is not an unquestionable good. An efficient system of racial profiling is worse than an inefficient one. An efficient deportation apparatus destroys more lives. An efficient welfare compliance system can manufacture another Robodebt before the public understands what has been built.
The question is never simply whether the software works. It is: what violence does the software allow the state to perform more efficiently?
Australia is surrendering digital sovereignty
Australia’s reliance on Palantir also raises a fundamental question of sovereignty.
Palantir is a United States company embedded in American defence, intelligence and immigration enforcement. Australian agencies are placing sensitive public functions inside systems controlled by a foreign corporation operating under foreign law.
The United Kingdom is beginning to confront this danger. Its parliamentary Science, Innovation and Technology Committee described Palantir’s growing public-sector role as an “unacceptable point of weakness” and urged the government to consider using a 2027 break clause in the company’s £330 million NHS contract. The committee warned that dependence on a small number of US technology providers could leave essential public infrastructure vulnerable to foreign interests.
Australia, meanwhile, continues to hand Palantir contracts.
The company has reportedly worked with Defence, AUSTRAC, Victoria’s Department of Justice, Coles, Rio Tinto and Westpac. Its technology is not sitting at the edge of Australian society. It is moving into institutions controlling money, prisons, national security, employment, resources and essential services.
Once a platform becomes embedded, replacing it becomes difficult and expensive. Staff are trained around it. Databases are structured through it. Decision-making processes come to depend on it. The contractor does not need to own the underlying data to exercise enormous power. Control over the infrastructure through which information becomes usable is power enough.
This is vendor lock-in as political capture.
Shut it down before it becomes irreversible
Australia needs more than another internal review or a privacy assessment released years after deployment.
There must be an immediate moratorium on new government contracts with Palantir and an independent public audit of every existing arrangement across Commonwealth and state agencies. That audit must identify what information is processed, where it is stored, who can access it, how datasets are combined, what automated inferences are generated and whether data can be accessed under foreign law.
The government must disclose whether any NDIS participant or provider information has been processed using Palantir technology, directly or through AUSTRAC or another Fraud Fusion Taskforce agency.
The Taskforce’s unredacted governance documents, information-sharing arrangements, privacy assessments and technology contracts must be released. Disabled people and their representative organisations must have decision-making power over what happens next, not token consultation after systems have already been installed.
There must also be enforceable prohibitions on using information collected for welfare, disability support, health or social services for policing, immigration enforcement or intelligence purposes without a warrant and strict judicial oversight. Australia must stop treating access to public support as consent to total state visibility.
Disabled people are not potential fraud events. Welfare recipients are not intelligence subjects. Migrants are not targets on a dashboard. Palestinians are not data points in a targeting system. People do not surrender their human rights because a government agency holds information about them.
Palantir’s mythology is that it allows powerful institutions to see everything. Our task is to make the institutions using it visible.
We need to know where Palantir is operating, which records it can touch and which decisions its systems shape. We need to know which agencies have merged information collected under entirely different legal and social purposes. We need to know what has already been built before it becomes so deeply embedded that governments tell us dismantling it is impossible.
After Robodebt, Australia promised that automated state cruelty would never be permitted to hide behind bureaucracy again. Yet the databases are connecting. The agencies are sharing. The contracts are multiplying. The safeguards arrive after the surveillance begins, if they arrive at all.
Palantir is not waiting outside the door. It is already inside the house. And the government will not tell us which rooms it can enter.
