OpenAI robotics lead quits after Pentagon AI network deal

Caitlin Kalinowski, a member of OpenAI’s robotics team, resigned after the company agreed to deploy its AI models on the Pentagon’s classified network.

In a post on X, she warned the deal could enable surveillance of Americans and lethal autonomous systems.

OpenAI confirmed her departure, saying the Defense Department deal allows national security AI use while banning domestic surveillance and autonomous weapons.

Sam Altman, CEO, said the company’s rush to reach a deal “looked opportunistic and sloppy.”

The agreement followed the breakdown of talks between the Trump administration and Anthropic, which had sought assurances against mass surveillance of Americans and fully autonomous weapons.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

The Pentagon’s move against Anthropic uses national security law in a new way

The Pentagon’s “supply-chain risk” label for Anthropic appears to be a new use of this power against a US-based AI company; the statutes and their legislative history focus on supply-chain threats tied to foreign adversaries 1.
The order tries a “secondary boycott” that would block any company working with the US military from also taking commercial work with Anthropic 1.
The move could hobble Anthropic by cutting off access to cloud providers such as Amazon or Google, since both also hold large US defense contracts 1.
Legal analysts say the rationale looks like a cover, with a national security tool used to punish a contract dispute rather than address a real security threat 1.

The OpenAI deal marks a tougher style of government pressure on AI companies

The dispute lays out a playbook that lets the US executive branch use national security tools to lean on AI firms during policy fights 1.
President Trump publicly called on “EVERY Federal Agency” to stop using Anthropic’s technology, which adds direct pressure 1.
This puts AI companies in a hard spot, with a choice to comply with government demands on disputed uses or face commercial penalties that could threaten operations 1.
A willingness to effectively blacklist a large AI company such as Anthropic could chill pushback from other firms over the ethical terms of military contracts 1.