Breaking

US Lifts Foreign-Access Curbs on Anthropic's New AI Models Amid Criticism of Secret Deal

July 1, 2026 at 10:16 EDT

Policy & Regulation
Foundation Models
Security

The US government lifted, in late June, the restrictions it had imposed on foreign-national access to Anthropic's latest AI models, Claude Fable 5 and Claude Mythos 5. But because the safety measures said to be the condition for lifting the curbs remain undisclosed, criticism of a "secret deal" between the government and the AI company is spreading.

June–July 2026 · Anthropic

A Frontier AI Model Was Blocked on National-Security Grounds — Then Quietly Cleared

Washington ordered all foreign-national access to Fable 5 and Mythos 5 cut off on June 12. Roughly three weeks later an undisclosed deal lifted the restriction — and the secrecy has fueled calls for clearer AI regulation.

Jun 12

Access-cut directive issued; both models disabled for all users

~3 wks

Restriction stayed in force until an agreement was reached ~Jun 30

Jul 1

Fable 5 redeployed globally; deal terms undisclosed

The Sequence

Jun 9 · Launch

Fable 5 & Mythos 5 announced after extensive red-teaming

→

Jun 12 · Block

Foreign-national access cut off after a safeguard workaround was reported

→

~Jun 30 · Deal

Undisclosed agreement endorsed by CAISI

→

Jul 1 · Restore

Fable 5 back online worldwide

Senior-Engineer Benchmark

Score out of 100 — Fable 5 leads on long-horizon, complex coding tasks

Fable 5

Opus 4.8

GPT-5.5

Praise

Strong on complex, long-horizon coding & agentic workflows
Broadly positive developer reception
Safety-first: defense-in-depth, tighter guardrails

Criticism

Slow, expensive and guardrails seen as too strict
Heavy token use — ill-suited to short tasks
30-day data retention adds customer burden

Why It Matters

A first case of ITAR-like national-security authority applied to a commercial AI model — settled behind closed doors.

If such measures spread industry-wide, new model rollouts could stall.

Anthropic proposes an industry jailbreak-severity scoring framework with Amazon, Microsoft & Google.

Transparency and fairness in AI regulation are now the central question.

Continue reading

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