LLM Digest
Subscribe

AI Storyline

6 items · 3 sources · 5 days

View as JSON

Operational story trace

Claude Fable

Current stateRedeployed · safeguards publishedstatus changed Jul 2

Latest change

Anthropic published details on Fable 5's cyber safeguards — what its classifiers do and don't block — plus a first-draft jailbreak severity framework, two days into the post-suspension redeploy.

Earlier contextThe story so far

Anthropic launched Claude Fable 5 and the unguarded Mythos 5 as long-horizon models "made safe for general use," but a US export-control directive citing national security suspended access days later — critics argued the controls themselves weakened US cyber defense. The models returned gated on Amazon Bedrock behind a data-sharing opt-in, then redeployed from July 1 once the controls lifted, backed by updated cybersecurity safeguards.

editor-curated · source-linked

State over time

● available · Jun 9–13safeguards published · Jul 1 → now ●
  • available · Jun 9–13
  • suspended · Jun 13–20
  • Bedrock w/ data-sharing · Jun 20–30
  • redeployed · Jun 30 – Jul 1
  • safeguards published · Jul 1 → now
LAUNCH · Jun 30
Claude Fable 5 + Mythos 5 ship — "made safe for general use"
1 source · show source ▾

Claude Fable 5 and Claude Mythos 5

anthropic_newsroomJun 30

The launch itself — Anthropic ships Fable 5 and the Mythos-class Mythos 5, billed as "made safe for general use."

THE TURN · Jun 15
US export-control directive suspends access
Citing national security, the US government orders access to Fable 5 and Mythos 5 suspended days after launch.
1 source · show source ▾
FRICTION · Jun 16
Critics say the controls themselves harm US cyber defense
1 source · scout · show source ▾
CONDITIONAL RETURN · Jun 20
Back on Bedrock — but only with data-sharing opted in
1 source · scout · watcher updated status · show source ▾
REDEPLOY · Jun 30
Export controls lifted — redeploying July 1 with new safeguards
1 source · watcher updated status · show source ▾

Redeploying Claude Fable 5

anthropic_newsroomJun 30

Export controls lifted — Anthropic redeploys Fable 5 starting July 1 with updated cybersecurity safeguards and a new industry jailbreak framework.

NOW · Jul 2
Anthropic details what the cyber classifiers block
1 source · scout · watcher updated status · show source ▾

What to watch — open questions

  • Does the July 1 redeployment restore standard private-inference terms, or does the Bedrock provider_data_share requirement persist?
  • Do the published cyber classifiers close the specific jailbreak that triggered the export-control suspension, or does exposure remain?
  • Will other frontier labs adopt the first-draft jailbreak severity framework, or does it stay Anthropic-only?
  • Is Fable 5 / Mythos 5 redeploying beyond Bedrock — first-party API or other clouds — on private-inference terms?
How this thread was built
scout surfaced 3editor wrote the arc · 6 beatswatcher 3 status changes

Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.