Q&A with Palantir CTO Shyam Sankar on the Iran school strike and AI targeting, the DOD dropping Anthropic, AI liability, effective altruism, and more
The following is the full transcript of the interview with Shyam Sankar, chief technology officer of Palantir, a portion of which aired on …
Context & Ripple Effects
Palantir’s public case for defense AI has been building through executive interviews: Sankar previously addressed surveillance and military AI, while Karp argued that reliance on AI labs for military technology was strategically unsound and promoted Palantir’s Nvidia work for U.S. agencies. The new interview moves that argument from capability and supplier strategy toward targeting accountability and liability.
The company’s executives have repeatedly used long-form interviews to defend Palantir’s role in politically contentious state technology, including Sankar’s earlier discussion of surveillance and war. A reported Pentagon break with Anthropic makes the terms on which AI vendors serve defense customers a more immediate commercial and governance question.
First-order effects
- Anthropic faces an immediate challenge to its defense-business position if the DOD ends the relationship, while the department must reassess which AI suppliers and deployment terms meet its requirements.
- Palantir gains a prominent venue to argue that targeting systems require identifiable operational accountability, rather than treating model providers as interchangeable infrastructure.
Second-order effects
- Other AI vendors seeking defense work face closer scrutiny of their deployment restrictions, liability posture, and willingness to support operational use cases.
- Defense buyers may place more value on contractors that combine models with data integration, audit trails, and accountable workflows, rather than procuring model access alone.
Third-order effects
- If supplier disputes over military use become recurring, defense AI procurement may separate into state-compatible providers and vendors that limit certain operational deployments.
- Liability for AI-assisted targeting is becoming a structural procurement issue: agencies and contractors will need clearer allocations of responsibility among model developers, system integrators, and operators.
The trend: Defense AI is shifting from a contest over model capability toward a contest over operational control, supplier alignment, and responsibility for real-world outcomes.