Breaking

Trump Calls Anthropic a Prior National Security Threat

June 19, 2026 at 14:22 EDT

Policy & Regulation
Foundation Models
Security

President Donald Trump said in an interview on "The Axios Show" that he had viewed AI company Anthropic as a "national security threat." Axios reported the remarks exclusively on June 19, 2026. Referring to the company and its CEO Dario Amodei, Trump said it was "not now" a threat but "a week ago, maybe" it was, and after a meeting at the G7 he described Amodei as a "nice guy, smart guy," adding that Anthropic had responded quickly and responsibly. He also suggested a competing company may have flagged the issue.

June 19, 2026 · Anthropic vs. Washington

Trump steps back: Anthropic "may have been" a security threat — but not now

A week after the Commerce Department imposed export controls that disabled its top models, the President signals partial de-escalation — calling CEO Dario Amodei "nice" and "smart," and ruling out shutting the company down.

models globally suspended to comply — Fable 5 & Mythos 5

1st

U.S. company ever labelled a "supply chain risk to national security"

ranking on FrontierCode at launch

Senior Engineer benchmark (Every.to, score /100)

The model at the center of the standoff also leads on coding evaluations.

Fable / Mythos 5

Opus 4.8

GPT-5.5

SWE-Bench Pro: 80.3% · Opus 4.8 69.2% · GPT-5.5 58.6%

How the standoff escalated

FEB 2026

Anthropic refuses Pentagon's military use, citing safety. Labelled a "supply chain risk"; agencies told to halt use.

→

JUN 12–13

After a reported Mythos jailbreak, Commerce imposes export controls; both models suspended worldwide.

→

JUN 19

Trump signals de-escalation: not a threat now, no plan to shut the company down.

Developers praised

"The best coding model"
Excellent for agentic & long-horizon tasks
1M-token context, Jan 2026 cutoff

Critics warned

"Expensive and slow," heavy on tokens
Guardrails trigger frequently
Jailbreak was narrow, not universal — banning it disarms defenders

The bigger picture

Washington is now treating frontier AI models like nuclear technology.

Export controls on a U.S. model mark a first. Even as the President eases the pressure, debate over spillover to firms like OpenAI — and broader uncertainty across the U.S. AI industry — is expected to continue.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.