Palantir’s Manifesto: What It Reveals About AI Vendors, Military AI and Procurement Risk

Palantir’s 22-point manifesto landed like a bomb: nationalist rhetoric, calls for a US military draft and blunt talk about AI-enabled weapons. For public-sector buyers, that’s not just PR — it’s procurement risk.

Key terms (quick definitions)

• Federated data platform: a system that links datasets across organisations without centralising them in one database, letting partners query shared records while keeping local control.
• AI agents: software systems that act autonomously to perform tasks, make decisions, or orchestrate other software and hardware on behalf of users.
• Dual-use: technology designed for civilian purposes that can also be adapted for military or surveillance applications.

What happened

Palantir published a 22-point manifesto on X (formerly Twitter) that mixes geopolitical ambition, cultural judgments and a practical acceptance of weaponised AI. Two lines from the text have particularly sharp edges:

“Some cultures have produced vital advances; others remain dysfunctional and regressive.”

“The question is not whether A.I. weapons will be built; it is who will build them and for what purpose.”

The manifesto echoes themes from Palantir CEO Alex Karp’s recent book and immediately prompted strong responses from UK MPs and civil society groups. That backlash matters because Palantir already holds more than £500m in UK public-sector work, reportedly including a roughly £330m role on an NHS federated data platform, contracts with the Ministry of Defence, and an engagement with the Financial Conduct Authority to analyse internal intelligence data.

Immediate fallout and reactions

Critics described the document as tone-deaf at best and dangerous at worst. MP Martin Wrigley called it:

Palantir’s statement “embraces AI state surveillance of citizens along with national service in the USA … a parody of a RoboCop film, or a disturbing narcissistic rant.”

Campaign group Foxglove’s Tim Squirrell said the tone was “comic-book villain worthy … fixated on US dominance and utterly unsuited to being anywhere near our public services.”

Palantir pushed back by pointing to outcomes it says its software is enabling in the UK — speeding NHS operations, shortening cancer-diagnosis times, supporting the Royal Navy and protecting vulnerable people — and emphasised that a substantial proportion of its workforce is UK-based. A company spokesperson said:

“Palantir software is helping to increase NHS operations, reduce the time it takes to diagnose cancer, keep Royal Navy ships at sea for longer, and protect women and children from domestic violence.”

Parliamentary questions and civil-society pressure have since pushed procurement teams and regulators to reassess risks tied to vendors with defence links and forceful public stances.

Why procurement teams should treat this as material risk

1) Reputational exposure: citizens notice who touches their data

A vendor’s public views can cost contracts, provoke parliamentary inquiries, and erode citizen trust. When a supplier signals a preference for militarised futures or cultural hierarchies, elected officials and watchdogs may pause or cancel live programmes — creating operational disruption for services that rely on those systems.

2) Data governance complexity: access isn’t just technical

Systems such as federated data platforms are designed to limit centralisation but still rely on vendor software for indexing, analytics and access control. Any vendor with broad analytic capability and deep infrastructure access becomes a de facto steward — and that stewardship raises questions about logging, auditability, third-party subcontracts, and the possibility of repurposing tools for surveillance or military ends.

3) Strategic alignment and dual-use risk

When AI agents can be adapted from civilian tasks to autonomous effects in a military context, procurement has to move beyond feature lists. Buyers must ask whether a supplier’s roadmap or intellectual property includes dual-use capabilities, and whether those capabilities are governed, auditable and constrained by contract and regulation.

Three short risk scenarios

• NHS screening paused: public outcry leads to a temporary halt on a federated analytics service while MPs demand audits, delaying critical cancer-screening rollouts.
• Regulatory analytics suspended: a financial regulator pauses a vendor’s analytics pipeline pending review, hampering market surveillance and requiring manual workarounds.
• Export-control trigger: an algorithm developed for logistics is found to be adaptable to unmanned weapon control, prompting an export-control investigation and supply-chain interruption.

Practical checklist for buyers

Treat supplier governance as a strategic item on the procurement agenda. Immediate actions procurement and C-suite leaders should take:

• Map precisely which systems and people can access personal or regulated data — and require vendor-supplied access logs.
• Insert a reputational-notification clause requiring the supplier to alert the buyer to public statements or publications that could materially harm trust.
• Demand dual-use disclosure: suppliers must list capabilities that are designed for or adaptable to military use and allow independent verification.
• Require independent audits and source-code or model access under NDA for high-risk components.
• Build a data escrow and exit ramp: define data return/erasure timelines, forensic log access and transitional support on termination.
• Set clear limits on subcontracting and foreign transfer of sensitive models or datasets.
• Embed KPIs tied to governance (e.g., percent of audited queries, timely breach notification) — not just uptime.
• Plan continuity: identify alternate providers and test transition procedures annually.

Sample contract clauses (drop-in language)

Reputational risk notification:

“Supplier must notify Buyer within 7 business days of any public statements, publications or material messaging by senior executives that could reasonably be expected to materially harm the public trust or reputation of Buyer. Buyer reserves the right to require remedial action, additional safeguards, or to suspend services pending review.”

Dual-use disclosure and audit:

“Supplier must disclose any AI capabilities designed for or adaptable to defence, surveillance or autonomous kinetic use. Buyer shall have the right to commission an independent third-party audit of such capabilities under NDA at Supplier’s expense.”

Data exit and forensic access:

“On termination, Supplier will return or irreversibly delete all Buyer personal and regulated data within 30 days, provide verified logs for the prior 24 months, and maintain forensic access for a further 90 days to support audits or investigations.”

Legal, regulatory and standards context

Procurement teams should stitch contractual protections to the legal environment. Key references:

• Data protection law (GDPR) — obligations for lawful processing, purpose limitation and data subject rights.
• UK Public Contracts Regulations — procurement fairness and transparency obligations.
• NIST AI Risk Management Framework and OECD AI Principles — practical frameworks for risk assessment and governance.
• Export controls and defence procurement rules — watch for dual-use red flags that may trigger export or national-security review.

Balance and counterpoints

Palantir’s defenders point to concrete results: reduced diagnostic latency, operational lifts in defence logistics, and assistance for safeguarding vulnerable people. Those outcomes matter. Removing valuable capabilities overnight can harm services and citizens.

But capability does not erase risk. Oversight and proportionality are the right responses: demand transparency and limits, not reflexive bans. Where a supplier provides unique benefits, buyers should haggle for governance — not just discounts.

Key takeaways

• What happened?
  Palantir published a 22-point manifesto on X that stirred political backlash while the company remains a major UK public-sector supplier.
• Why it matters for procurement?
  Vendor politics can create real operational, reputational and regulatory disruptions for buyers who rely on their systems.
• What to do next?
  Treat vendor ideology and dual-use potential as contractable risks: require disclosure, audits, exit plans and reputational-notification clauses.

Questions procurement leaders should ask right now

• Who at the vendor can access our data?
  Demand a named list of accounts, roles and third-party subcontracts with access rights and audited logs.
• Does the supplier have dual-use capabilities?
  Require written disclosure and independent verification for any functionality that could be repurposed for defence or surveillance.
• What happens if public trust collapses?
  Insist on a tested exit plan, escrowed data and transitional support to keep services running without vendor lock-in.
• Are there regulatory or export risks?
  Run legal review for GDPR, procurement rules and potential export-control triggers before signing or renewing contracts.

If your organisation buys AI technology, assume politics is a line item. Start by demanding dual-use disclosures, independent audit rights and a robust exit plan — then negotiate the governance you need before capability alone decides your supplier relationships.