Fetching from the wire…
Public story · 2026-07-02 · high
A new classifier blocks the vulnerability-surfacing bypass over 99% of the time, capping the model at half of weekly usage through July 7.
Why now: Anthropic confirmed the redeployment on July 1, ending three weeks of uncertainty that began with the June 12 export order.
Anthropic put Claude Fable 5 back into general use on July 1, nearly three weeks after a US export-control order pulled it offline over a safety bypass, per Anthropic.
For Pro, Max and Team subscribers, that means Fable 5 counts against usage again, up to 50% of weekly limits through July 7. For Anthropic, it's a test of whether a frontier model can stay available once a government export order flags it as a risk.
The order followed an Amazon report on Fable 5's safeguards. Amazon found they could be tricked into surfacing software vulnerabilities.
Anthropic says a new classifier blocks that specific technique in more than 99% of cases.
This confirms what had circulated since June 26, when unconfirmed reports said Fable 5 was on its way back. What's different now is the specificity: a classifier trained against the exact bypass Amazon found, not a broad safety overhaul.
A 99% catch rate is Anthropic grading its own test. The 50% cap through July 7 reads like a trial run before full reinstatement, not a victory lap. If a red-teamer finds a way past the classifier before July 7, watch for the usage cap to tighten rather than lift. That's the tell the fix isn't holding.
The same dynamic, a model's availability swinging on a government order, is what makes LongCat-2.0's export-free training worth watching.
Each link below shares sources, entities, or timing with this story.
Anthropic published "Redeploying Fable 5" on July 18, and the headline reads like good news until you get to the metering. Fable 5 becomes a permanent subscription feature for Max and Team Premium. Good. But it's metered at 50% of standard weekly limits, meaning every Fable to...
Day 9 of the blackout, and the reporting finally caught up to the politics. According to Tom's Hardware, the export-control shutdown of Claude Fable 5 and Mythos 5 wasn't a cold bureaucratic move. White House AI adviser David Sacks says the administration asked Dario Amodei to...
On June 9, Anthropic released Claude Fable 5 and Mythos 5 across Claude.ai, Claude Code (CLI and web), and Cowork. The spec sheet: 1M-token context, 128K max output, a January 2026 knowledge cutoff, and pricing at $10 input / $50 output per million tokens. That's double Opus 4...
New reporting (June 17-18, via WIRED/Washington Post coverage) details the standoff behind the June 12 Claude Fable 5 / Mythos 5 suspension: SK Telecom's access was revoked, then an Amazon-filed vulnerability report triggered the global export ban. The administration, via Davi...
Anthropic's leaders sat down with Trump administration officials on June 15. They left with nothing. Claude Fable 5 and Mythos 5 stay offline for foreign nationals under the Commerce Department's export-control directive, and an administration official called restoring G7 acce...
Moonshot AI released Kimi K3, a sparse mixture-of-experts activating 16 of 896 experts per token. That's about 1.8% of the pool live at any moment, with a 1M-token context window and native vision. Two new architectural pieces show up: Kimi Delta Attention and Attention Residu...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.