Fable 5 is back. Now catch the next drop first.
claude-fable-5 came back on July 1, 2026 — the US export controls were lifted and Anthropic redeployed it globally. FableWatch did its job. Want to be first to the next frontier drop? ArtificialWatch watches every major model — Claude Opus 5 reportedly next, already spotted on Google Vertex AI — and pings you free the moment one launches or goes dark.
✳ FableWatch is an independent project — it is not Anthropic, and is not affiliated with, endorsed by, or operated by Anthropic.
What happened to Fable 5, in plain terms.
Anthropic launched claude-fable-5 (Fable 5) on June 9, 2026 for Claude Max buyers.
On June 12, 2026, Anthropic suspended access after a US Commerce Department export-control directive — reportedly a letter from Secretary Howard Lutnick to CEO Dario Amodei.
As of July 31, 2026, Fable 5 is back — Anthropic redeployed it globally on July 1, 2026 after the export controls were lifted, with a new safety classifier that Anthropic says routes some cybersecurity requests to Opus 4.8. FableWatch flipped to online the moment it answered.
FableWatch is an independent monitor — not affiliated with Anthropic — that caught the return the minute it happened. It now pings the model endpoint every 60 seconds and watches Anthropic's lineup for the next Mythos-class model (Fable 6, or 5.1), alerting subscribers the moment a new state-of-the-art model is live.
It's not just an alert. It's the whole drop, put to work.
The next Mythos-class drop won't come with a warning. Four things fire the moment the monitor trips.
you get told, loudly
Email & push fire instantly the moment it lands, and SMS + an automated call reach verified numbers the same minute. On The Swarm, an optional 12-hour Swarm Alert re-pings your own verified phone until you acknowledge — so a meeting or a night's sleep doesn't mean you miss it.
your own VPS, running on the new model
We spin up a dedicated VPS for you and run dozens of API calls in parallel through a 400+ question set on your own Anthropic key— the answers appear on your dashboard as they're captured (refresh to update).
GitHub fixes, drafted for you
Connect your GitHub and, when the next model lands, the Swarm reads a PR diff or an issue on your repos and drafts a suggested fix as a comment for you to review.
one place for all of it
The alert, the captured answers, and the suggested-fix comments all land in a single dashboard (refresh to update). Stop the swarm, browse the answers, open the comments.
Pick how ready you want to be.
One-time, not subscriptions. Start free; upgrade any time.
✳ Fable 5 shipped Jul 1 — so FableWatch now watches for the next Mythos-class model (Fable 6, or 5.1). When it drops, the free list hears in a batch — The Signal hears first.
1,861 watching- One email when a new state-of-the-art model is released
- Sent in the batch — after paid members are alerted
- No card. Unsubscribe anytime.
- No SMS or phone call
- Not first in line — paid members are alerted before you
- Notified FIRST when a new state-of-the-art model drops — ahead of the entire free list
- SMS + an automated phone call (after you verify your number)
- Every channel you enable fires the same minute it lands
- Priority queue: your alert goes out before free emails
- A dedicated VPS, spun up on the new state-of-the-art model the moment it's available
- Dozens of parallel calls run a 400+ question set on your own Anthropic key — answers appear on your dashboard as they're captured (refresh to update)
- GitHub fix suggestions: the Swarm drafts a suggested fix as a comment on your connected repos & issues
- An optional 12-hour Swarm Alert that re-pings your own verified phone until you acknowledge
- Top-priority SMS, call & email — first in line, ahead of everyone
- Everything in The Signal, plus a dashboard for the whole run
- Bring your own Anthropic API key — you control the spend cap
We launch a VPS for you, and point it straight at the new model.
The moment it drops, your phone gets the Swarm Alert — opt-in texts and calls that repeat until you acknowledge — while we spin up a dedicated VPS that runs dozens of API calls to the new model in parallel through a 400+ question set on your own key, connects your GitHub to draft suggested fixes on your PRs and issues, and streams it all to your dashboard. Bring your own Anthropic key — you set the spend cap. The extraction run and fix-comment drafts happen only after we manually confirm the drop and you approve the run — never automatically.
From launch to working for you, the moment it drops.
The Swarm turns the next model's launch into provisioned compute, a streamed answer set, and drafted fix comments — on your own key, once we confirm the drop.
- 01
The next model lands
Our monitor catches it live and answering. It trips — once, debounced, no false alarms.
- 02
We launch your VPS
A dedicated server spins up for you in seconds — pre-armed with the 400+ question set and your own Anthropic key.
- 03
Dozens of API calls run in parallel
Workers query the new model all at once, working through the question set on your key and capturing each answer as it lands.
- 04
Your GitHub gets a suggested fix
Connect a repo and, the moment it lands, the Swarm reads a PR diff or an issue and drafts a suggested fix as a comment for you to review.
- 05
It all lands on your dashboard
The answer set and the drafted fix comments appear on your dashboard as they're captured (refresh to update).
A 412+ question evaluation set spanning reasoning, coding, long-context, and safety probes — built for this monitor. A category-spanning sample:
parseConfig()comment draftedDon't scramble. Queue your prompts now.
Line up the prompts you'll run the moment the next model answers — saved in your browser, free. The Swarm can run the whole list for you on your own key once we confirm the drop.
Before you decide, the honest answers.
Is claude-fable-5 back yet?
Yes — Fable 5 is back. FableWatch's monitor confirmed claude-fable-5 answering on July 1, 2026 at 7:29 PM ET and alerted the watch. Per Anthropic, it returned with a new safety classifier that routes some cybersecurity-adjacent requests to Opus 4.8. The live status at the top of this page tracks it continuously — and FableWatch now watches for the next Mythos-class model.
Why was claude-fable-5 suspended?
Anthropic suspended access to claude-fable-5 on June 12, 2026, reportedly after a US Commerce Department export-control directive — described in reporting as a letter from Secretary Howard Lutnick to Anthropic CEO Dario Amodei. The specifics are attributed to news outlets and remain partly unconfirmed; FableWatch is not affiliated with Anthropic and does not have inside information.
What is FableWatch watching now?
The next Mythos-class model. Fable 5 was the first Mythos-class release, and the next one — Fable 6, or a 5.1 — is the drop worth being first to. FableWatch pings the claude-fable-5 endpoint every 60 seconds and watches Anthropic's model lineup for new Mythos-class ids; the moment a new state-of-the-art model is live, subscribers are alerted. Our guarantee still stands: if you bought a paid tier and its alert never fired within 90 days of purchase, email us and we'll refund you in full. You're paying for the window, not a gamble.
Is this affiliated with Anthropic?
No. FableWatch isn't affiliated with, endorsed by, or connected to Anthropic. We built the independent monitor that caught claude-fable-5's July 1 return the minute it happened — and now we watch for the next drop.
Is it a subscription?
No. Every paid tier is a single one-time charge — The Signal and The Swarm both. Nothing recurring, nothing to cancel.
What exactly do I get for paying?
The Signal: you're alerted first — email and push fire instantly the minute a new state-of-the-art model answers, ahead of the free list, with SMS and an automated call to your verified number the same minute. The Swarm: all of that, plus a dedicated VPS that runs a 400+ question set in parallel on the new model on your own Anthropic key, plus GitHub fix suggestions — the Swarm drafts a suggested fix as a comment on your connected repos — with every answer appearing on your dashboard as it's captured (refresh to update).
When do SMS and calls actually fire?
Email and push are instant. SMS and an automated call are live too — US carrier registration (A2P 10DLC) has cleared, so once you verify your number, texts and calls fire the same minute Fable answers. And per our guarantee, if a paid tier never reaches you we refund it in full — you're never out money for a channel that couldn't deliver.
Fable's back. Here's the full toolkit.
Which Anthropic model fits which job now that claude-fable-5 is live again — the closest fit for each task, and honest about the offensive-cyber capability only the restricted Mythos tier is meant to fill.
The suspension, logged.
A sourced timeline of the claude-fable-5suspension — why Anthropic disabled Fable 5 on June 12, 2026, and how it came back on July 1. Newest first, attributed to its source and hedged where it isn't settled. Not affiliated with Anthropic.
- Jul 17:29 PM ET
It's back: Claude Fable 5 is answering again — confirmed live at 7:29 PM ET
It's real this time. At 7:29 PM ET on July 1, FableWatch's monitor confirmed claude-fable-5 answering on the official Anthropic API — seven consecutive clean probes, debounced — flipped the status to ONLINE, and fired the return alert to the watch. We independently double-checked with a direct API call, and Fable 5 returned a normal completion (a plain “PONG” to a one-word ping). This is the genuine restoration, not a repeat of the June 25 selector/UI mirage: the model is serving live traffic again. It lands one day after Anthropic's June 30 “Redeploying Fable 5” post set a July 1 global return, and 19 days after the June 12 export-control directive took Fable 5 and Mythos 5 offline. One honest asterisk carries over from Anthropic's own post — a new safety classifier now routes some cybersecurity-adjacent coding and debugging requests to Opus 4.8, so a fraction of prompts won't run on Fable while it's tuned. But the question this entire timeline was built around — “is it actually back?” — is, as of tonight, yes.
- Jun 30PM ET
Anthropic makes it official: Fable 5 returns globally tomorrow — with a new classifier that routes some cyber work to Opus 4.8
Anthropic put its own name on the resolution. In a June 30 blog post — “Redeploying Fable 5” — and a companion @AnthropicAI thread, the company confirmed Fable 5 will be “available again globally tomorrow,” Wednesday, July 1, on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork, with AWS, Google Cloud, and Microsoft Foundry to be re-enabled “as quickly as possible.” On pricing, for Pro, Max, Team, and select Enterprise plans it's included for up to 50% of weekly usage limits through July 7, then moves to usage credits — which squares with the earlier datamined “usage credits” reporting. Mythos 5 was restored to a set of US organizations after the government's June 26 approval, with wider domestic and international Glasswing access still being coordinated. The catch worth flagging up front: the return ships with a new set of classifiers that Anthropic says “target and block more cybersecurity tasks,” so in the near term some routine coding and debugging will be blocked and fall back to Opus 4.8 — users are told when a request is redirected — and Anthropic says it will refine the classifiers over the coming weeks to cut false positives. One honest caveat holds: by Anthropic's own word access begins “tomorrow,” so the model isn't actually answering yet — the legal block is gone and the date is set, but FableWatch still flips only when it truly responds.
- Jun 30PM ET
The post-mortem in the blog: the Amazon report, a “>99%” classifier fix, and a cross-industry jailbreak framework
The same post doubles as Anthropic's post-mortem — and its pitch for what comes next. On the trigger: the June 12 export directive followed a report in which Amazon researchers bypassed Fable 5's safeguards to make it identify software vulnerabilities and, in one case, produce code demonstrating how one could be exploited. Anthropic's framing is that this exposed nothing unique — it says its own testing found less-capable models (Opus 4.8, GPT-5.5, Kimi K2.7) could identify the same vulnerabilities, that every model it tested could reproduce the single exploit demo, and that the technique reached no unique Mythos-level capability — a “borderline” safeguard case over routine defensive work. Its fix: a retrained classifier it says now blocks that specific technique in “over 99%” of cases, which Commerce's Center for AI Standards and Innovation (CAISI) — per Anthropic — tested against both the old and new safeguards and judged “extraordinarily strong.” Looking forward, Anthropic says it's drafting a consensus framework with Amazon, Microsoft, Google, and other Glasswing partners to score jailbreak severity on four axes — capability gain, breadth, ease of weaponization, and discoverability — backed by a 24/7 monitoring team and a new HackerOne program for Fable 5 cyber jailbreaks, plus deeper US-government collaboration (pre-release access to models and safeguards, misuse information-sharing, and dedicated joint research) building on the June 2 Executive Order. Read the specifics as largely self-reported — Anthropic's own capability tests and its own “>99%” and “extraordinarily strong” characterizations — and note the company still concedes it's “probably impossible” to make any model fully robust to jailbreaks.
- Jun 30PM ET
It's official: Commerce withdraws the Fable 5 and Mythos 5 export controls — Anthropic restores access tomorrow
It's official, and it's both models. In a June 30 letter to Anthropic's Tom Brown, Commerce Secretary Howard Lutnick withdrew the controls from his June 12 order: “A license is no longer required for the export, reexport, or in-country transfer, including deemed export or deemed reexport, of the Mythos or Fable models.” The reversal follows commitments from Anthropic — to proactively detect and address security risks, to work with the U.S. government on protocols, standards, and releases for Mythos, Fable, and future models, and to report any malicious activity — together with the Bureau of Industry and Security's reassessment of the diversion risk. Anthropic confirmed within the hour: “We've received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5. We'll begin restoring access tomorrow, and will share an update soon.” Two caveats keep this honest. First, access isn't live this second — by Anthropic's own word, restoration begins tomorrow, so the model isn't answering yet even though the legal block is gone. Second, Lutnick explicitly “reserves the right to reevaluate… and the necessity of reimposing a license requirement, should circumstances change or should Anthropic fail to adhere to its commitments.” But after 18 days, this is the resolution the whole saga pointed to: the export controls on Fable 5 and Mythos 5 are gone. The watch now shifts from “will they lift it?” to “is it actually back?” — and that part is hours, not weeks, away.
- Jun 30PM ET
POLITICO: Trump administration expected to lift Fable 5 export controls — as soon as tonight
The strongest signal yet that the suspension is ending. POLITICO's Sophia Cai and Cheyenne Haslett — Cai broke much of the original Fable/Mythos story — report (June 30, ~7 PM ET) that the Trump administration plans to remove the export restrictions on Anthropic's Fable 5 “as early as Tuesday evening,” citing two people familiar: a senior White House official and an administration official, both granted anonymity to discuss internal planning. The change would let Anthropic make Fable 5 available again to all general users — the first time since the June 12 controls that consumers, not just vetted Mythos partners, would be back in scope. It follows Friday's partial Mythos lift (Jun 26), which pointedly left Fable blocked, and lands as the administration tries to balance frontier-model cyber risk against industry pressure (POLITICO notes OpenAI likewise agreed last week to curb access to its ChatGPT-5.6 model under White House pressure). The caveat that matters: this is a reported, expected action — not a done deal, and not a restoration you can use yet. It's two anonymous officials describing internal planning that could still slip, and this saga has already produced at least one false “Fable is back.” If the order issues tonight, US users and others should regain access; until it actually lands and the model answers, it's the most concrete report so far — not the finish line. We'll know within hours.
- Jun 30PM ET
App-string leak: Fable 5's return looks gated behind paid credits — and identity verification
A fresh datamine of the Claude app hints at how Fable 5 will actually be sold when it returns — and ties it to the ID checks Anthropic had framed as unrelated. Leakers M1Astra (@M1Astra) and Jimmy Apples (@apples_jimmy), via the devmode app-datamining community, surfaced new strings building out a credit-purchase flow: “Buy usage credits to use Fable 5,” “Fable 5 runs on usage credits, billed separately from your plan,” and — the notable one — “Your credits will be added once your identity is verified.” A companion “Purchase under review / Your credits will be added once the review is complete” state implies purchases can be held for compliance screening. Two things stand out. First, the “billed separately from your plan” framing cuts against the earlier Claude Code read that Fable would fold into Pro/Max subscription limits — pointing instead to metered, pay-as-you-go credits. Second, it puts identity verification directly in the Fable purchase path, after Anthropic had publicly cast its new ID checks as a general account measure not specific to Fable. The hard caveat: these are datamined strings, not an announcement — unreleased UI copy can change, ship gated, or never ship, and the credit-to-ID coupling is the leakers' read of code, not stated policy. Still, it lines up with the broader reported picture (usage-credit pricing at API rates; Persona-run ID and facial verification slated around July 8 per Decrypt, TechTimes), and it's the most concrete look yet at what re-entry to Fable 5 may require: a verified identity and a separate wallet.
- Jun 27PM ET
Axios: Fable 5 could be back next week — if the Pentagon and NSA sign off
The first concrete timeline for Fable's return — with a hard condition attached. Axios reports (June 27) that Fable 5 could come back as soon as next week, but only if the Pentagon and the NSA both sign off. The sticking point is at Defense: Secretary Pete Hegseth still has Anthropic formally designated a “Supply-Chain Risk to National Security,” a label that has to be lifted or worked around before consumer access resumes. So the picture now has two gates: Commerce has started loosening its export controls (the Mythos carve-outs in Friday's letter), but Fable 5 for the public still waits on a national-security sign-off from the Pentagon and NSA. Read it as the most specific “back” signal yet — a possible next-week window — paired with the clearest remaining blocker. Nothing is official until both clear it, and “could” is doing real work in that sentence.
- Jun 27PM ET
The letter's fine print: who Friday's Mythos lift actually covers — and Fable 5 isn't in it
A closer read of Commerce Secretary Howard Lutnick's letter to Anthropic's Tom Brown shows just how narrow the partial lift is. The Mythos 5 license requirement is waived for three groups only: Anthropic's own non-US-citizen researchers; designated US “trusted partners” and their foreign-national employees; and US-government users — civilian agencies and the national labs. Everyone else still needs an export license to touch the model. Two things did not move: Fable 5 remains fully banned, with no carve-out at all, and every requirement from the original June 12 letter stays in effect — including the threat of “prompt criminal and civil penalties” for violations. Anthropic, for its part, is described as committed to working with the US government on protocols, standards, and future releases. Lutnick also kept his leverage explicit: “I reserve the right to reevaluate and adjust the scope of license requirements… should circumstances change.” A gated loosening for vetted Mythos users — not a reopening, and nothing yet for the consumers waiting on Fable.
- Jun 26PM ET
First official rollback: US clears Mythos 5 for 100+ trusted partners as a broader deal nears
The biggest move since the suspension — and the first official one. In a Friday letter to Anthropic chief compute officer Tom Brown, Commerce Secretary Howard Lutnick wrote: “I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model,” citing “significant progress” in the daily talks. The government has cleared Mythos 5 for 100+ US institutions — companies and government agencies — a gated, partial lift, not a public reopening (Semafor, Bloomberg, 9to5Mac). Separately, Bloomberg reports the two sides are “moving closer” to an agreement that would lift the curbs on both models, which “could be lifted once officials across the administration give their blessing.” The big caveat: this is Mythos for vetted partners — Fable 5 for consumers is still down, with no public restoration announced. But after two weeks of standoff, it's the first concrete loosening from Commerce itself, and the clearest signal yet that a full lift is within reach.
- Jun 25AM ET
“Fable 5 is back” — debunked: Anthropic says it's serving zero traffic
Viral “Fable 5 is back, with video proof” claims spread after the model reappeared in app/model selectors and some users got responses — but Anthropic staff knocked it down. Sam McAllister (Anthropic): “We are currently serving exactly 0 traffic to Fable 5. This could be a UI bug though, will track it down.” Head of Growth Amol Avasare likewise called the access reports false — a front-end bug or misinformation (corroborated by TestingCatalog, Alex Volkov, and an X Community Note). Selecting Fable still returns the “currently unavailable” notice pointing to Anthropic's suspension page. So the selector sightings are a UI artifact, not a restoration — Fable 5 and Mythos 5 remain suspended with no traffic served. A reminder that “it shows up in the dropdown” isn't the same as “it's back.”
- Jun 24PM ET
Signs of a return: Claude Code strings — and a Bedrock relisting — point to Fable 5 coming back
Concrete prep signals are stacking up, though still short of an official return. Claude Code v2.1.190 quietly added the string “You've used your included Fable 5 usage for this week” and dropped “purchased separately from your plan,” pointing to Fable 5 returning bundled into Pro/Max/Team subscriptions with weekly usage limits rather than as a paid add-on (Decrypt). Separately, observers report Fable 5 has reappeared in the Amazon Bedrock catalog/runtime, with Mythos 5 and Mythos Preview on the control plane but not yet runtime — the banned models quietly resurfacing in AWS infrastructure (observer-reported, not an official AWS notice). None of this is a restoration: no Commerce sign-off, no Anthropic statement, and the export controls still stand. But it's the strongest infrastructure-level evidence yet that a return is being staged.
- Jun 24PM ET
WIRED: Anthropic benched Amodei from White House talks — Tom Brown now leads
WIRED reports a telling shift in the negotiations: Anthropic has sidelined CEO Dario Amodei from the White House meetings about Fable 5, sending cofounder Tom Brown instead — and the administration is happier for it. A White House insider's blunt version: “Tom Brown is not being a weirdo like Dario and can actually engage.” Brown and Amodei reportedly clashed over how to handle the jailbreak fallout. The change lines up with the visible thaw — by June 19 Trump was calling Anthropic “very responsible” — suggesting the logjam was as much about rapport as technical terms. It reframes the path back as partly relationship repair, not just a safeguards fix.
- Jun 21PM ET
Senator relays NSA claim: Mythos broke into “almost all” classified systems “in hours”
The Economist reported remarks from Sen. Mark Warner (vice chair of the Senate Intelligence Committee), who said Gen. Joshua Rudd — head of the NSA and Cyber Command — told him that on June 11, Mythos broke into “almost all of our classified systems, not in weeks, but in hours.” Read it carefully: this describes a red-team exercise — Mythos pitted against replicas of classified environments, finding and chaining vulnerabilities far faster than a human team — not a confirmed breach of live NSA systems, and at least one industry figure has publicly disputed even the framing. Still, it's the most dramatic rationale yet for the speed of the June 12 shutdown, reframing the trigger from “a narrow jailbreak” to “a model that out-hacks the government's own teams.” Treat the specifics as a senator's secondhand account via a single outlet, not an established fact.
- Jun 21PM ET
The concrete path back: ID checks and facial verification from July 8
One thread worth pinning down amid the optimism: how US users would actually regain access. Because Anthropic can't verify a user's nationality in real time, its updated privacy policy adds identity verification — a government-issued ID plus a face photo or video and what it calls “facial geometry templates” (which Anthropic concedes “may be considered biometric data in some jurisdictions”) — slated to begin around July 8 (TechTimes, CIO). The aim: confirm US citizenship so verified Americans could use Fable 5 even if the foreign-national ban stays in force. It's a workaround, not a resolution — the export controls remain in place, and it trades a compliance problem for a privacy one (handing over biometric ID to use a chatbot), which privacy advocates have flagged. Notable as the first concrete, dated mechanism for the models' partial return.
- Jun 21PM ET
Rumor: a more capable Mythos successor has reportedly finished training
AI observer Andrew Curran said on June 21 that “a new, more capable version of Mythos has emerged from training” — possibly to be called Mythos 5.1 or Mythos 6, or kept internal to accelerate development (echoed by beincrypto and others). Anthropic has not confirmed the model's existence, name, or capabilities, so treat this as an unverified report from a single well-followed watcher. If true, it sharpens a point critics keep making about the export controls: pulling Fable 5 and Mythos 5 from public access does nothing to slow the underlying research — and may even free up resources — while open rivals keep shipping. A rumor, but a telling one. (Separately: a recycled “Sonnet 5 / Fennec” leak made the rounds this week, but that surfaced back in February and reportedly already shipped as Sonnet 4.6 — not a new development.)
- Jun 20PM ET
White House and Anthropic start drafting a joint AI-risk “rulebook”
Multiple outlets — Politico, Yahoo Finance, TheStreet, Tekedia — report that the White House and Anthropic have begun drafting a joint risk framework: a shared standard for judging how severe an AI security flaw is and when the government should step in, meant to replace ad-hoc emergency interventions in future frontier-model incidents. Negotiators are said to be defining how far safeguards were bypassed, what capabilities were exposed, and the real-world impact of a breach. It's the first constructive, forward-looking thread in the saga — but it's a process, not a resolution: the framework is still being drafted, it's aimed at future incidents rather than this one, and Fable 5 and Mythos 5 remain suspended with no restoration announced. A path toward a playbook, not the ban lifting.
- Jun 19AM ET
Trump softens: Anthropic “not now” a national-security threat
In an interview for “The Axios Show” (June 19), President Trump markedly softened his posture. Asked whether he views Anthropic — or CEO Dario Amodei — as a national-security threat, he said “Well, not now, but a week ago, maybe,” and praised the company's conduct: Amodei “responded very responsibly, I thought,” and “so far it's been very responsible.” He came away from the G7 calling Amodei “nice” and “smart.” It's the warmest signal yet from the very top — but it's a shift in tone, not a resolution: the export controls remain in place, no restoration of Fable 5 or Mythos 5 has been announced, and Trump pointedly did not rule out invoking emergency Defense Production Act powers if Anthropic doesn't “get in line.” Read it as de-escalation, not the all-clear.
- Jun 18PM ET
Bipartisan House members demand Lutnick justify the controls
A bipartisan group of House members sent Commerce Secretary Howard Lutnick a letter Thursday pressing him to explain the export controls on Fable 5 and Mythos 5 (Washington Post). They want the specific statutory authority he invoked to block foreign nationals, the legal and technical rationale, and whether rival AI companies should expect the same treatment — warning the order may set “a precedent with significant implications” for the industry. It's the first pushback from Congress, separate from the cybersecurity-executive and legal-scholar criticism already on the record. The administration's posture is unchanged (Sacks: remediate the jailbreak, then the controls lift), and as of now the models remain suspended with no finalized deal — Anthropic's “coming days” confidence still untested.
- Jun 18~AM ET
Anthropic: “very confident” models return “in the coming days”
Anthropic's Managing Director of International, Chris Ciauri, told a press conference in Seoul: “We are very confident that in the coming days, the models will become available again” (Korea JoongAng Daily). It's the most optimistic on-record signal since the suspension — but it's stated confidence during ongoing talks, not a finalized deal or a verified restoration date. Separately, the Washington Post reported a new origin detail: a Korean telecom company with access to Claude Mythos may have triggered the directive due to suspected ties to China — a more specific national-security thread than the Amazon-jailbreak angle. Prediction markets ticked up on the optimism but still aren't pricing an imminent return. Holding both: the tone from Anthropic is now “days away,” while the WIRED-reported condition (“guarantee no jailbreaks”) remains technically contested.
- Jun 17~2–5 PM ET
The catch: White House wants “no jailbreaks,” experts say that's impossible
WIRED reported the administration's actual condition for restoring Fable 5: Anthropic must ensure the model's guardrails can't be circumvented — and security experts say that can't be done for any model. Katie Moussouris (the one outside expert who read the vulnerability report) detailed the technique publicly: defensive prompts like “fix this code” plus manual steps to generate test scripts, which she argues can't be removed without making the model worse at the legitimate defensive security work it's meant for — and which apply equally to every capable model, including foreign and open-weight ones US export controls can't touch. She called the directive “misguided.” Separately, legal experts (The Verge) questioned the mechanism itself: export controls traditionally govern things that cross borders, and this appears to be the first time they've been applied to using an AI model remotely — raising the question of what Anthropic is even “exporting.” The takeaway across coverage: the government's stated bar for return may be technically impossible to meet, which could mean a longer standoff than the “going fine” optimism suggested.
- Jun 17~1 PM ET
Trump: negotiations “going fine” — but markets stay skeptical near-term
Trump commented on the talks for the first time, at the G7 in France: asked about the Anthropic negotiations, he said “Going fine, I think it's going fine,” and Lutnick (present) echoed “Going fine” (WSJ). The remarks came at a G7 lunch with AI executives including Dario Amodei; WSJ reports Altman and Hassabis were seated beside Trump while Amodei sat across the table next to Macron. Prediction markets aren't betting on a quick fix, though: Polymarket's “restored for US customers by…” market sits at ~39% for July 1 (~22% by June 22, ~30% by June 26), and a separate “rescinds the foreigner ban by…” market is ~27% for June 30. Trump's tone is the most encouraging signal since the suspension began, but the money still leans against a near-term return.
- Jun 17morning ET
First government-side rationale surfaces (Reuters)
Reuters, citing a copy of the letter, reported Lutnick's stated reason for the order: officials feared Fable 5 and Mythos 5 could be deployed by military-intelligence users in China, Russia, or other “countries of concern.” The letter ordered suspension of export “to destinations worldwide and all foreign nationals, wherever located.” This is the first government-side rationale on record — until now everything came through Anthropic's framing. The legal basis is reportedly the Export Controls Reform Act of 2018 (dual-use provision). Commerce did not respond to questions about the specific authorities. (Note: the stated concern is potential adversary access — distinct from the still-uncorroborated claim that a Chinese group actually accessed the model.)
- Jun 16now
Still suspended — Anthropic disputes the “refused to fix it” claim
No resolution, no return date. Anthropic's side now disputes the “refused to fix it” framing — a source close to the company says it was never presented with details and never refused (Fox Business). The claim that a Chinese group accessed the model remains uncorroborated. No CVE assigned, no public technical disclosure, no formal Commerce statement. The story is in slow negotiation mode.
- Jun 16~12 AM ET
Jailbreak reportedly just “fix this code”
Overnight reporting (Fortune, Tom's Hardware, Fox Business, CNBC): the jailbreak technique is reportedly as simple as “fix this code,” per Katie Moussouris of Luta Security, whom Anthropic asked to review the report. The Amazon / Jassy path is now confirmed across outlets, no longer WSJ-only. Trump adviser David Sacks says Anthropic “refused” to fix it before the controls came down, with the administration framing it as Anthropic's “recklessness” and a test case for a new AI-guardrails executive order. A claim that a Chinese group accessed the model is flagged unconfirmed. Status: still suspended, conditional restoration on the table.
- Jun 15late PM ET
Monday's talks end without lifting the controls (WIRED)
Monday's talks concluded without restoring access; next steps unclear. The administration still believes Fable 5 can be jailbroken to reach Mythos-level capability; Anthropic reiterated the concerns are overblown. Lutnick dialed in from the G7; Cairncross didn't participate himself. Commerce signaled willingness to restore Fable 5 for consumer use only, once Anthropic resolves the jailbreak concern — a conditional, pending-a-fix hold rather than an indefinite ban.
- Jun 158:25 PM ET
Resolution likely to take longer than a few days (ABC)
A senior White House official said resolution would take longer than a few days (“that's up to Anthropic”). Trump asked Lutnick to get more involved. Saturday saw hour-plus calls between Lutnick, Cairncross, and Anthropic's Tom Brown and Sarah Heck. Anthropic assembled a technical team — Logan Graham (Frontier Red Team), Dave Orr (safeguards), Nicholas Carlini (security) — for Monday's in-person presentation to ONCD and Commerce's CAISI (Chris Fall present). Treasury sat it out.
- Jun 153:20 PM ET
Anthropic flies senior staff to DC (NY Post)
Anthropic flew senior technical staff to DC over the weekend; an in-person Commerce meeting was set for Monday. Intel reportedly came from Amazon CEO Andy Jassy directly; a senior official said “nearly half a dozen” companies raised concerns, not just Amazon. 40+ cybersecurity executives — including Alex Stamos, plus people from Adobe and Zoom — signed a letter to Lutnick opposing the controls. Separately, Anthropic is already suing the administration over a Pentagon supply-chain-risk designation.
- Jun 13~6:30 AM ET
Still suspended; AWS confirms revocation
No resolution. AWS/Bedrock confirmed the revocation on their end; Commerce stayed silent. Added context: Anthropic's ~$965B confidential IPO filing. The Amazon angle was still single-sourced at this point.
- Jun 132:47 AM ET
WSJ: Amazon researchers reported the jailbreak
The Wall Street Journal reported the jailbreak was found by Amazon researchers, who reported it to Commerce; Axios reported the administration had already tried to get Anthropic to delay launch. Open questions: why Amazon — Anthropic's biggest investor — was the one to report it, and why it went to Commerce rather than direct disclosure. (WSJ-sourced.)
- Jun 12evening
Anthropic disables Fable 5 and Mythos 5
Unable to filter foreign nationals in real time, Anthropic disabled Fable 5 and Mythos 5 for all customers; AWS/Bedrock revoked access too. Other models — Opus 4.8, Sonnet, Haiku — were unaffected. Anthropic published a statement disputing the action, calling it a likely misunderstanding and saying it was working to restore access.
- Jun 125:21 PM ET
Commerce orders suspension
Anthropic received an export-control directive from the Commerce Department — a letter from Secretary Howard Lutnick to Dario Amodei — citing national-security authorities and ordering suspension of access for any foreign national inside or outside the US, including Anthropic's own foreign-national employees. The letter did not spell out the specific concern.
- Jun 92026
Anthropic launches Fable 5 and Mythos 5
Anthropic launched Fable 5 and Mythos 5 as its most capable public models. Fable 5 is the publicly-available version with safeguards; Mythos 5 is the underlying cyber-capable base model, restricted to limited access.
How we know, and how we hedge.
The status and the timeline run on the same rule: we measure what we can verify, and we attribute the rest. Not affiliated with Anthropic.
The monitor
We poll the claude-fable-5 endpoint every 60seconds — it caught the July 1 return the minute it happened — and watch Anthropic's model lineup for new Mythos-class ids. An event is confirmed only on a debounced successful response — not a single blip — so a brief flicker can't fire a false alert.
The timeline
Every claim in the suspension timeline is attributed inline to the originating outlet — reporting we've drawn from outlets including WSJ, WIRED, ABC, Fortune, Fox Business, and CNBC, among others — and marked reportedly or unconfirmed until it's corroborated. We never state an unverified claim as fact.
ArtificialWatch
Future AI model availability + restoration notices, by email only when you explicitly opt in. Separate opt-in. No SMS. No payment.