Hello everyone,

Nineteen days ago, the world's most capable AI model vanished from the internet. Not because it broke. Because the U.S. government decided it was too dangerous to let anyone outside the country touch it.

That's not hyperbole. It's the actual reason Claude Fable 5 disappeared from Claude.ai, the API, and Claude Code on June 12. This week, it came back. But it didn't come back the same.

On July 1, Anthropic restored global access to Fable 5 across the Claude Platform, Claude.ai, Claude Code, and Claude Cowork. The move followed a June 30 decision by the Department of Commerce to lift the export control order it issued just three days after Fable 5's June 9 launch.

The order never targeted Fable 5's intelligence. It targeted its passport. Because Anthropic couldn't verify user nationality in real time, it had to pull the model for everyone, everywhere, rather than risk violating a directive built around foreign-national access. Mythos 5, Fable's less-restrained sibling, was already back in the hands of vetted Project Glasswing partners by June 26 — a preview of the broader thaw that arrived this week.

The trigger was a specific finding. Amazon researchers discovered a way to prompt Fable 5 into identifying software vulnerabilities, and in one case, generating exploit code. Anthropic pushed back hard on the severity, noting that Opus 4.8, GPT-5.5, and other models could surface the same flaws. It still built a new safety classifier to close the gap. Per Anthropic, that classifier now blocks the reported technique in over 99% of cases.

The tradeoff: more legitimate coding and debugging requests are getting flagged as suspicious and rerouted to Opus 4.8 instead. Anthropic has acknowledged this openly and says it's still tuning the balance.

If you're on Pro, Max, or Team, mark your calendar.

The real story isn't that Fable 5 came back. It's what Anthropic agreed to so it could. The company committed to giving U.S. government evaluators early access to future models that meaningfully push the capability frontier on national-security-relevant work, before the public ever sees them.

That's narrower than a blanket AI review board, but it's a meaningful precedent for the industry. And it raises the question every frontier lab is now facing: how much runway do you lose when a regulator gets a seat at your model's launch table?

Most SaaS finance teams know their billing process isn't perfect. Few know what it's actually costing them.

Answer 5 quick questions — contracts signed per month, ACV, days to first invoice, error rate, DSO — and the Tabs Billing Lag Calculator gives you a dollar figure benchmarked against top SaaS companies.

It takes two minutes. The number might surprise you.

Calculate your billing lag and see where you stand.

Calculate your billing lag

If your team leans on Fable 5 for security research, red-teaming, or anything touching vulnerability discovery, budget for more friction. Expect occasional handoffs to Opus 4.8, and don't assume a decline means your prompt was malicious. It might just be the new classifier being cautious.

Longer term, watch this precedent closely. Pre-release government vetting could become a moat, not just a compliance step. Only labs with the infrastructure to run parallel government evaluations will be able to ship at the frontier. That's a very different competitive landscape than move fast and figure out safety later.

How will pre-release government vetting change how much you trust the next frontier model? Reply and tell us — we read every response.

See you next time.