Palantir's Sankar Calls Bloomberg Strike Report 'Wrong'
TL;DR
- Sankar declared Bloomberg's reporting 'wrong' on air, attributing the Minab school deaths to 'human error' before the Pentagon's final investigation report has been released.
- BBC sources confirmed Claude remained inside Maven for active Iran operations as recently as the week before the Pentagon's Oct. 5 completion announcement.
- Anthropic's removal was driven by CEO Dario Amodei's refusal to strip safety guardrails against autonomous weapons, a values-based disqualification rather than a capability gap.
The Palantir CTO went on television and called the Bloomberg story wrong.
Shyam Sankar, interviewed by Margaret Brennan on CBS's Face the Nation, was asked about reporting that tied a strike which killed schoolchildren to over-reliance on Palantir's AI tooling. "The Bloomberg reporting is wrong. Yeah," he said. He attributed the incident to human error, declined to go further pending a final report, and said the company has since "built additional AI agents that did more checking."
On the Pentagon's move to stop using Anthropic's Claude, he was flatter. "No impact." The Department of Defense, he told Brennan, already draws on "a wide range of model providers" including xAI, OpenAI, and NVIDIA's Nemotron, casting institutional sovereignty as a function of a crowded model menu rather than any single vendor relationship.
He saved his sharpest line for the AI-safety camp. Effective altruism, he told Brennan, is "really this millenarian doomsday cult meets messianic complex." The doomsayers, in his telling, "have basically been wrong" on job displacement; economists, he said, "see that actually we're having job growth."
Then the figure he came to deliver. "The ability to execute 1000 strikes in a single day would have taken two weeks of planning by 50 people before, and being able to do that with one person in four hours provides you a significant battlefield advantage." He named no product, no operation, and no customer.
What others are reporting
-
BBC News Read →
Reveals Claude was still operational inside Maven in Iran operations the week before the Oct. 5 announcement, exposing the gap between official phase-out claims and operational reality.
BBC sources said Claude remained in military use as recently as last week, including in operations involving Iran.
-
Benzinga Read →
Covers Sankar's CBS interview directly, foregrounding the 'human error' framing and Palantir's post-incident deployment of additional AI checking agents as a defensive disclosure.
-
Yahoo Finance Read →
Adds the market dimension: PLTR hit a record high the day of Sankar's interview, with retail and institutional investors aligned behind the rebuttal.
"The Bloomberg reporting is wrong," Sankar said, adding that "the likely cause is human error."
-
Forkast News Read →
Frames Anthropic's exit as a values-based disqualification rather than a capability gap, centering the safety-architecture tradeoff that determined who remains in the military AI stack.
"Once embedded, the removal process is inherently painful," regarding Anthropic's integration into Palantir's Maven Smart System.
-
Foreign Policy Journal Read →
Frames the simultaneous reputational defense and bullish analyst upgrades as a single market event, showing financial and geopolitical narratives converging on the same day.
"The Bloomberg reporting is wrong," Sankar said, adding that "the likely cause is human error."
-
Sunday Guardian Live Read →
Places Sankar's interview inside the ongoing Iran-US war, noting the investigation's conclusion could shape military procurement and international rules of engagement for AI in active conflict.
"The likely cause is human error," Sankar told Brennan, arguing responsibility for verifying targets rests with human commanders rather than automated software.
Shared on Bluesky by 1 AI expert
Originally reported by cbsnews.com
Read the original article →Original headline: Palantir CTO Sankar Disputes Bloomberg Iran Strike Report on Face the Nation, Shrugs Off DOD Dropping Anthropic