FABLE 5 IS BACK ONLINE! BUT THERE IS A LESSON HERE

Anthropic's Fable 5 is back. The model came online for users on Wednesday after the Trump administration lifted the export controls that had forced a suspension barely a week after its June 9 debut. The brief blockade created a mess for Anthropic, which had no way to verify users' nationality and pulled the plug on both Fable 5 and the Mythos 5 model rather than risk violating the rules. Now the government has reversed course, and the most capable model Anthropic has ever shipped is live again.

WHY IT WAS BLOCKED IN THE FIRST PLACE

The controls were triggered by a specific security finding. Amazon researchers discovered a prompt that could bypass Fable's safeguards, and the model itself was able to identify software vulnerabilities. Anthropic investigated and found that older Claude models and models from competing companies could also locate the same vulnerabilities. The issue was not unique to Fable 5, in other words, but the model's power made it the focus of concern.

you may like

ANTHROPIC’S NEW FABLE 5 MIGHT BE SILETNTLY NERFING YOU

OPENAI RELEASE NEW MODELS, LIMITS ACCESS AT GOVERNMENTS REQUEST

MOONSHOT'S KIMI K3 MATCHES TOP US MODELS AT FRACTION OF THE PRICE

Anthropic responded by shipping a new classifier that blocks the bypass technique in more than 99 per cent of cases. That fix, combined with deeper coordination with the government, appears to have been enough to get the controls lifted. Anthropic says it is now deepening cooperation with the U.S. government on pre-release testing, information sharing, and research collaboration. The message is clear: a model this capable is not getting released again without the government knowing exactly what is in it.

WHAT FABLE 5 CAN ACTUALLY DO

Fable 5 is not a small step up from previous Claude models. It outperforms the Opus line on longer, complex tasks and can work autonomously for longer than any prior Claude model. It is the most powerful publicly available AI tool, and Anthropic is positioning it as a capability leap rather than an incremental improvement.

That extra autonomy and reliability comes with a cost. Fable burns through tokens faster than other models, and Anthropic has structured access to match the demand. For now, until July 7, subscribers on Pro, Max, Team, and select Enterprise plans can use up to 50 per cent of their plan's weekly usage allotment on Fable 5. After that cap is hit, additional Fable use requires paying for tokens outside the subscription. After July 7, Fable will be available only through usage credits, no longer included in any plan.

Customers who want heavy Fable usage will essentially need to commit to a metered approach. And even then, queries that Anthropic deems to pose security or safety risks may be silently routed to less powerful models. The company is not taking chances after the Amazon finding.

THE PRECEDENT THIS SETS

Fable 5's brief grounding is the most visible example yet of the Trump administration asserting direct control over a frontier AI release. The government applied export controls after the model's ability to find software vulnerabilities was identified, using the controls to stop foreign nationals from accessing it.

The move raises significant questions about how and when the administration will block future model releases. OpenAI has already said it is withholding broad release of GPT 5.6 while it consults with the government, which requested the delay. If the pattern holds, every major frontier model from now on may face a waiting period while the government decides whether to allow foreign access. That is a structural shift for the industry, not a one-off event.

THE BIGGER QUESTION

Anthropic restored access to Mythos 5 for U.S. organisations that are part of Project Glasswing, a partnership for deploying AI in national security contexts. That suggests the company is already mapping out a two-tier access model: fully open for domestic use, restricted or banned for foreign users when the government decides the model is too powerful.

The bigger question is whether Fable 5's release is the last time Anthropic ships a model without a pre-negotiated government clearance. The deepening cooperation with the government on pre-release testing points toward a model where Anthropic and the administration effectively co-sign any future frontier release. That would be a major departure from the industry's previous approach of releasing first and dealing with regulators later.

what to read next

NUCLEAR IS BACK IN EUROPE AND AI IS THE REASON

MICROSOFT BLOCKS FABLE 5 FOR EMPLOYEES

FRENCH GOVT MESSAGING PLATFORM TCHAP BREACHED, 14GB STOLEN

For now, Fable 5 is live, available to anyone willing to pay per token or use their weekly allocation. But the controls that blocked it for a week are not going away. They are just moving from a blunt suspension to a more permanent oversight framework. The genie is out of the bottle, but the government now has a hand on the cork.