Anthropic CEO Dario Amodei is back in talks with Pentagon after refusing to drop AI safeguards

Anthropic has been in the limelight for refusing an AI deal with the US Defence Department due to AI reliability and readiness for mass surveillance and autonomous weapons. Now, CEO Dario Amodei is back to reach an agreement with the US Defence Department, after risking potential exclusion from the military's supply chain. 

The company is reportedly in talks with Emil Michael, the under-secretary of defence for research and engineering, over a deal that would define how the Pentagon can use Anthropic's AI models, as per a Financial Times report.

If the deal is finalised, the US military can continue using Anthropic's AI models and tools, and the company would likely avoid the damaging "supply chain risk" designation that has been hanging over it. Last week, Defence Secretary Pete Hegseth threatened to apply that label to Anthropic, though it has not yet been made official.

Anthropic drew a line last week on how its AI tools will be used by the US military. The company refused to remove two safeguards from its contract with the Pentagon, firstly, a ban on using its models for mass domestic surveillance and a restriction on deploying fully autonomous weapons systems.

Amodei, in a blog post, said that AI is not reliable enough to be trusted without human oversight. Now, the company is reportedly back to negotiations with the Pentagon. However, both sides have not reached an official deal as of now, since we are still waiting for confirmation.

While Anthropic’s refusal to the US military could risk the company's status in the US, its Claude AI chatbot gained much popularity among users. On the other hand, OpenAI faced major backlash for striking a deal with the Pentagon. It also started a “Cancel ChatGPT” across social media platforms, and it led to a 295% surge in uninstallation last week.