NYC Hospitals End Palantir AI Contract Amid Privacy Concerns – UBOS Tech News

NYC Health + Hospitals has decided not to renew its contract with Palantir, ending a $4 million AI analytics agreement that was slated to expire in October 2026.

The decision, announced in March 2026, follows intense activist pressure, mounting privacy concerns, and parallel scrutiny of Palantir’s £330 million deal with the United Kingdom’s National Health Service (NHS). For healthcare administrators and technology journalists, the move signals a pivotal shift in how public‑sector health systems evaluate AI analytics providers and safeguard patient data.

Why NYC Health + Hospitals is ending the Palantir contract

During a New York City Council hearing, Dr. Mitchell Katz, President of NYC Health + Hospitals, explained that the Palantir agreement was always intended as a short‑term pilot focused on revenue‑cycle optimization. The contract, signed in November 2023, has generated roughly $4 million in fees for Palantir, primarily to help the system recover Medicaid and other public‑benefit reimbursements.

Key reasons for non‑renewal include:

• Growing public and staff concerns that de‑identified patient data could be repurposed beyond research.
• Pressure from activist groups demanding a “firewall” against any data sharing with U.S. immigration authorities.
• A strategic shift toward in‑house data platforms that keep all health information under direct hospital control.

In an email to the press, the agency emphasized that after October, “no data will be shared with Palantir or used in any of the company’s applications.” The transition will rely on internally built tools that align with the UBOS platform overview, which promises end‑to‑end data sovereignty for public‑sector clients.

What Palantir’s AI platform does for hospitals

Palantir’s flagship product, Foundry, aggregates disparate data sources—electronic health records, billing systems, and social determinants of health—into a single analytical environment. In the NYC context, the platform was used to:

1. Identify gaps between services delivered and charges captured.
2. Automate claim validation for Medicaid, Medicare, and other public programs.
3. Generate “de‑identified” datasets that could be leveraged for secondary research, provided the city granted permission.

While Palantir asserts that it never owns customer data, critics argue that the company’s extensive integration capabilities could enable re‑identification when combined with other public datasets—a risk amplified by recent advances in generative AI.

Activist pressure and privacy concerns

The “Purge Palantir” campaign, coordinated by groups such as the American Friends Service Committee, the national nurses union, and BDS activists, mobilized protests, petitions, and a public‑records request that exposed the contract’s language on data de‑identification. A representative from the campaign said:

“We don’t think the same AI systems that target immigrants for ICE raids should be used in hospitals.”

Legal scholars, including Sharona Hoffman of Case Western Reserve University, warned that “de‑identification is no longer a guarantee; AI makes re‑identification increasingly feasible.” Their concerns echo similar arguments raised in the UK over the NHS contract.

Parallel scrutiny of Palantir’s £330 million NHS deal

Across the Atlantic, the UK’s National Health Service has signed a £330 million agreement with Palantir to deploy the same Foundry platform nationwide. The contract has sparked a separate wave of criticism:

• Health‑justice charity Medact warned that the system could enable “data‑driven state abuses of power,” including potential ICE‑style raids.
• MPs from the Liberal Democrats have called for a parliamentary inquiry into the deal, citing insufficient safeguards for patient privacy.
• Even the UK’s Financial Conduct Authority has faced backlash after awarding Palantir a contract to analyze internal intelligence data for financial‑crime detection.

The UK controversy underscores a broader global debate: should public‑sector health agencies entrust massive, proprietary AI platforms with sensitive patient information?

Reactions from officials, experts, and patient groups

NYC officials maintain that the contract was a “temporary revenue‑cycle tool” and stress that the new in‑house solution will retain the same analytical capabilities without external exposure.

Palantir’s spokesperson responded:

“Palantir, as a software company, does not own or have any rights to customer data – and each customer environment is individually protected against unauthorized access or misuse via robust security controls which can be fully administered and audited by the customer.”

Privacy experts remain skeptical. Ari Ezra Waldman of UC Irvine warned that any contract allowing “purposes other than research” effectively hands the government a back‑door to repurpose health data.

Patient advocacy groups such as the New York Health Coalition praised the decision, calling it “a victory for patient autonomy and data dignity.”

Implications for data governance in public health

The NYC withdrawal sets a precedent for how municipal health systems might approach public‑sector AI procurement:

• Contractual clarity: Future agreements will likely include stricter clauses limiting secondary data use.
• In‑house alternatives: Platforms like the Enterprise AI platform by UBOS are gaining traction as they promise full data control.
• Transparency dashboards: Public reporting of AI‑driven decisions will become a regulatory expectation.
• Cross‑border policy alignment: The parallel UK debate may push international standards for health‑data AI.

For administrators, the key takeaway is that “technology fit” must be evaluated alongside “ethical fit.” The rise of generative AI tools—such as the UBOS templates for quick start—offers a way to prototype internal solutions without exposing data to third‑party vendors.

Future outlook and call to action

As public health agencies worldwide grapple with the promise and peril of AI, several trends are emerging:

1. Hybrid governance models: Combining open‑source AI stacks with proprietary security layers.
2. Patient‑centric data trusts: Legal entities that hold data on behalf of patients, granting controlled access to analytics platforms.
3. Regulatory sandboxes: Government‑run environments where AI tools can be tested under strict privacy safeguards.

Healthcare leaders are urged to:

• Audit existing AI contracts for clauses that permit secondary data use.
• Invest in internal data‑engine platforms—such as the Workflow automation studio—to reduce reliance on external vendors.
• Engage with patient advocacy groups early in the procurement process.

The NYC decision demonstrates that public pressure, combined with clear policy alternatives, can reshape the AI landscape in health care. As the sector moves forward, the balance between innovative analytics and robust privacy safeguards will define the next generation of public‑sector AI.

For the original reporting, see the Guardian report.

Explore related UBOS solutions

If you’re evaluating alternatives to third‑party AI platforms, consider these UBOS offerings that align with modern data‑privacy standards:

• AI marketing agents – customizable bots that respect data boundaries.
• AI Article Copywriter – generate compliance documentation without exposing patient data.
• AI SEO Analyzer – optimize health‑service portals while keeping analytics in‑house.
• AI Video Generator – create patient education content securely.
• AI Chatbot template – deploy conversational assistants that run on your own servers.

The departure of Palantir from NYC’s public hospitals is more than a contract ending—it is a signal that data privacy, ethical AI, and local control are becoming non‑negotiable pillars of modern healthcare technology.