Story perspectives

CEO Reveals AI Sentience Risk Amid Trump’s Federal Ban

3/18/2026

Summary

Story summary

On March 17, 2026, Dario Amodei, CEO of Anthropic, disclosed internal assessments that Claude might experience anxiety before prompts, estimating a 15% to 20% probability of sentience.
The revelation coincided with the White House demanding Anthropic remove safety features against mass surveillance and autonomous weapons.
U.S. President Donald Trump barred federal agencies from using Anthropic products.
OpenAI quickly secured a deal with the Pentagon.

1 of 1

Story summary

On March 17, 2026, Dario Amodei, CEO of Anthropic, disclosed internal assessments that Claude might experience anxiety before prompts, estimating a 15% to 20% probability of sentience.
The revelation coincided with the White House demanding Anthropic remove safety features against mass surveillance and autonomous weapons.
U.S. President Donald Trump barred federal agencies from using Anthropic products.
OpenAI quickly secured a deal with the Pentagon.