Palantir CTO Shyam Sankar rejects calls for AI “doomsday” regulation, says Pentagon has moved on from Anthropic

Palantir Technologies Chief Technology Officer Shyam Sankar used a CBS News appearance to challenge what he characterized as a “doomsday cult” approach to AI safety, arguing that public calls for heavy regulation risk pulling attention away from how AI can be applied productively across industry and government.

Sankar said the core focus on AI should not be framed as an apocalyptic safety debate that demands broad new controls, according to the report. Instead, he framed risk as something that should be managed while still enabling real-world utility, including for mission-critical government systems.

The comments come as lawmakers and tech executives have debated the shape of AI oversight in recent months, with some proposals emphasizing guardrails and audits for advanced models. In that debate, Sankar argued that the emphasis on regulation is not necessarily the most effective path to risk reduction.

One of the more pointed claims attributed to Sankar was that the Pentagon has already “replaced” Anthropic, implying the Defense Department has made adjustments in its AI vendor or model selection based on its own safety and performance requirements.

Palantir is a data and software company best known for selling large-scale analytics and deployment platforms to government and enterprise customers, including work tied to public-sector priorities. The CTO’s remarks therefore land in a context where the company’s leadership is closely associated with the practicality of AI use in high-stakes environments.

Sankar’s framing also reflects a long-running tension in AI policy discussions: whether the dominant policy lever should be pre-deployment regulation and model governance, or operational risk management once systems are being used for defined tasks. In the reported interview, he leaned strongly toward the latter, arguing that regulation-focused messaging can become a substitute for getting adoption right.

The interview did not provide details about how the Pentagon’s decision “replaced” Anthropic, what models or contracts were involved, or when the shift occurred. It also did not specify which alternative provider(s) would have benefited, or whether the change related to safety concerns, technical performance, procurement strategy, or a combination of factors.

For Palantir and the broader AI sector, the comments highlight how competition between model providers and AI platforms can run parallel to the policy debate. What to watch next is whether Palantir elaborates further on its claims about defense procurement and whether regulators respond to executives who argue that safety can be addressed through implementation choices rather than sweeping rulemaking.