Breaking

US Government Lets Select Organizations Regain Access to Anthropic's Top Model

June 26, 2026 at 20:28 EDT

Foundation Models
Policy & Regulation
AI Agents

After weeks of negotiations, the Trump administration has permitted Anthropic to restore access to its most advanced AI model, Claude Mythos 5, for more than 100 trusted US companies and government agencies that operate and defend critical infrastructure. According to WIRED, the limited reinstatement came after talks led through the White House and Commerce Secretary Howard Lutnick.

June 26, 2026 · Commerce Department · Anthropic

White House Partially Lifts Ban on Anthropic's Most Advanced AI Model

After a sweeping export-control directive forced Anthropic to disable both new "Claude 5" models, regulators now allow restricted "Mythos 5" access to be restored for 100+ trusted US organizations — while the public "Fable 5" stays locked.

100+

trusted US organizations now eligible for Mythos 5 access

~2 wks

of Washington negotiations between full ban and partial easing

public model (Fable 5) still fully restricted, no return date

Three weeks, three reversals

The path from launch to ban to partial restoration

JUN 9

Mythos 5 & Fable 5 launched — first "Mythos-class" tier above Opus

→

JUN 12

Export control imposed — both models disabled for ALL users

→

JUN 26

Mythos 5 partially restored for approved US partners

Benchmark leap over Opus 4.8

New "5" models vs the previous Opus class — taller column = higher score

SWE-Bench Pro

80.3%

Mythos / Fable 5

69.2%

Opus 4.8

Humanity's Last Exam (with tools)

64.5%

Mythos / Fable 5

57.9%

Opus 4.8

Pricing: $10 / million input tokens · $50 / million output tokens

The case for

Major gains in agentic coding & long-running tasks
Called Anthropic's "strongest cybersecurity model"
Redeployed to approved cyber defenders & infrastructure

The concerns

Enterprise confusion from on-and-off restrictions
High pricing & guardrails triggering on cyber queries
Fears of a de facto government licensing regime

Wider ripple effect

OpenAI delayed GPT-5.6 at the government's request — signaling that advanced-model rollouts are now increasingly subject to regulators' judgment.

Continue reading

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