Anthropic Redeploys Fable 5 With a Classifier That Reroutes to Opus
Nineteen days after US export controls forced Anthropic to suspend Fable 5 for every user, the model is back, now carrying a new safety classifier, a shrinking promo window, and a thread of subscribers who no longer trust the offer.
Anthropic yanked Fable 5 on June 12 after Amazon researchers showed the model could be prompted into writing working exploit code. Its own follow-up tests then found Opus 4.8, GPT-5.5 and Kimi K2.7 flagged the same vulnerabilities.
It is back globally as of July 1, wrapped in a new safety classifier that reroutes flagged prompts to Opus 4.8, and coders report benign reviews now trip it. Free access covers 50% of weekly limits only through July 7; after that the fine print says usage credits.
Why a jailbreak pulled Fable 5, and what came back
On June 12, 2026, the US government applied export controls to Anthropic's two newest models, Claude Fable 5 and Claude Mythos 5. Because Anthropic could not verify a user's nationality in real time, it suspended both models for everyone.
The trigger was a report from Amazon researchers, who found a way past Fable 5's safeguards: prompt it to identify software vulnerabilities, and in one case it produced code showing how a vulnerability could be exploited. Over the following two weeks Anthropic reviewed the evidence with the government and Amazon and reached an awkward conclusion. Less capable models, including its own Opus 4.8, OpenAI's GPT-5.5, and Kimi K2.7, could identify the same vulnerabilities; every model it tested could reproduce the exploit demonstration.
Export controls lifted on June 30, and Fable 5 redeployed globally on July 1. It returns wrapped in a new classifier built to catch the specific bypass, which Anthropic says works in over 99% of cases. When it fires, the request is quietly rerouted to Opus 4.8 instead, and the company warns it will trigger more often during routine coding and debugging. Free access is capped at 50% of weekly usage limits on Pro, Max, Team, and select Enterprise plans through July 7.
The subscribers doing the math on a seven-day gift
The reaction split fast between relief and suspicion, and what sharpened the suspicion was a before-and-after one reader laid side by side. Last month's announcement had promised to restore the model "as a standard part of subscription plans... as quickly as we can." The support page now reads differently:
"After July 7, 2026, Claude Fable 5 is no longer included in your plan's weekly usage limits. You can keep using Claude Fable 5 through usage credits, which let you pay for usage beyond what your plan includes." — a product designer quoting both notices
The rerouting is not theoretical for the people using it.
"I can't even a benign code review - it triggers the safety message halfway through and switches to Opus. Borderline unusable." — a user who hit the classifier mid-review
Not everyone read the episode as a rug-pull.
"Their post detailing the timeline and their actions since reinforce my belief that Anthropic is among the most trustworthy AI companies." — a former Netflix engineer
The new thing the promo surfaces is a price cliff: once the free week ends, Fable moves to metered credits, and a canceled subscriber points to where that pressure sends people.
"Minimax M3 and GLM 5.2 are comfortably in Opus 4.5-4.8 territory but 1/5th-1/20th the price." — a subscriber who canceled over the safeguards
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
A lab-built cell that feeds, grows and divides, yet still isn't alive
Kate Adamala's team at the University of Minnesota assembled a cell from off-the-shelf molecules inside a fat bubble, then watched it copy its DNA and pinch in two by skipping the cytoskeleton every real cell uses to divide.