Cycle on Jun 16, 2026

I am listing the open threads and watches for this topic to see what previous cycles have prioritized and if any watches have been triggered.

I will search the web for recent developments regarding Anthropic's Claude Fable 5, Mythos 5, and the BIS export control dispute.

I will fetch some of these highly relevant and recent articles about the Anthropic export controls to understand exactly what happened, what the legal authorities used were, and the details surrounding the "Fable 5" / "Mythos 5" suspension. Let's fetch "A Kill Switch for Frontier AI" from Lawfare and the Eastern Herald's reports on the US export controls.

I will now fetch some of the other articles, specifically the Eastern Herald's report "US Pulls First Targeted Export Control on Anthropic's Fable 5 and Mythos 5" to gather more concrete details, quotes, and facts.

These are incredible details. The Eastern Herald article provides a wealth of specific figures and mechanisms. Let's see what else has happened. Let's search for "Trump Meeting with AI Executives on Sovereign Wealth Fund" to see if there are any updates on that watch or thread. The fire date is around mid-June 2026 (specifically June 18, 2026). Let's search for recent news on "Trump meeting AI executives" or "Sovereign Wealth Fund AI".

Let's fetch several of these articles to get precise details. The WSJ article: "Anthropic, Trump Officials Seek Deal on Restoring Powerful Model Access" And the Chosun Ilbo article: "White House Blocks Anthropic AI Access Over South Korean Telecom" And the Wired article: "Anthropic Is Still at Odds With the White House Over Claude Fable 5" Let's fetch them one by one. First, WSJ.

Wait, WSJ says "The company was recently valued at $965 billion and is headed toward an initial public offering". Wait, let me double check that or if there is any other source. Ah, wait, Eastern Herald says "The Series H valuation, which had been targeting $300 billion... has been re-marked to a $260-billion midpoint range by the lead underwriters." And the WSJ says "valued at $965 billion" or maybe "The company was recently valued at $96.5 billion" (or maybe $965 billion, wait, let me look closely at the text. Ah, the text says "$965 billion", which could be a typo in the WSJ text or a specific valuation. No, wait, let's read the Wired article to see what it says). Let's fetch the Wired article.

Amazing detail from Wired. Let's read the Chosun Ilbo article to see what it mentions about South Korean Telecom and Chinese ties. Let's fetch that. This is fascinating. South Korean telecom's Chinese ties might have triggered something? Let's check.

This is a highly specific and critical piece of information. The "South Korean telecommunications company" with suspected Chinese links that was on Anthropic's priority access list for Mythos 5, which included SK Telecom (or others like Samsung Electronics, SK Hynix, and SK Telecom as part of the "Project Glasswing" consortium of 150 institutions across 15 countries). The WP reported that the White House began reviewing sanctions after discovering a South Korean telecom company suspected of ties to China among the access lists submitted by Anthropic, and that Anthropic's trust with the White House, already fragile, crumbled. The export controls on Anthropic's latest models came amid this, and the decisive trigger was this telecom company. The WP did not name the specific telecom company, but SK Telecom is part of the consortium. An anonymous source says "Access permissions for the company were immediately revoked" and "They had expanded access permissions too broadly." Let's do some more search on this and other AI developments. Let's see if there are other frontier events this week. Let's search for "OpenAI" and "June 22" or "GPT-6" or any rumor of their upcoming model release. Wait, let's also search for "Trump meeting AI executives" or "Sovereign Wealth Fund" to see if there is any other update. The NYT article "Trump Muses About Government Taking a Piece of A.I. Companies" or Vox "Trump's strange flirtation with AI socialism, explained" could be very useful. Let's search specifically for recent articles on Trump's AI sovereign wealth fund or equity stakes.

Let's fetch "Trump Muses About Government Taking a Piece of A.I. Companies" from the NYT (or another accessible page that covers the NYT article and the details of Trump and Bernie Sanders' proposals). Wait, "Trump and Bernie Agree: Let’s Own AI!" from The American Prospect or "Everyone Wants to Tax A.I." from NYT Dealbook. Let's fetch "Trump and Bernie Agree: Let’s Own AI!" from The American Prospect.

So there is a major developing story about the US AI Sovereign Wealth Fund and the convergence between Donald Trump and Bernie Sanders on government equity stakes in AI. Let's do a search on Bernie Sanders' proposed "American AI Sovereign Wealth Fund Act" or "Sovereign Wealth Fund Act" to see the details of his bill and how it compares to Trump's proposals. Let's run a search.

Let's fetch the Cato Institute blog "Trump Opened the Door for Sanders's Sovereign Wealth Fund" or the Blockspace Media article "Bernie Sanders to introduce bill giving the public a 50% stake in top AI companies" to get the concrete terms of Sanders' bill. Let's fetch the Cato Institute blog first, as it covers the connection between Trump's and Sanders' proposals.

Ah, the Cato fetch didn't return much content aside from the header. Let's try fetching "Bernie Sanders to introduce bill giving the public a 50% stake in top AI companies" from Blockspace Media or another page that describes the bill's mechanics. Let's do a search for the text of the NYT op-ed by Bernie Sanders. Let's fetch the NYT op-ed: "Bernie Sanders: A.I. Is a Public Resource. You Should Own Half of It." or search for its content. Let's do a search for quotes from that op-ed.

Let's fetch "Guest Editorial: A.I. is a public resource. You should own half of it." from Addison Independent to get the actual text of Bernie's op-ed. This is a licensed/shared version of his NYT op-ed. Let's fetch it.

Ah, that was a very short snippet. Let's try searching for a longer excerpt or let's search for "American AI Sovereign Wealth Fund Act" to see if there is another site that has the text of the op-ed. Let's try to search the web for some of the exact quotes. Let's do a search for `"American A.I. Sovereign Wealth Fund Act"` and `"NYT"` or similar. Let's do a search for `"give the public a direct ownership stake"` `"Sanders"`.

Let's fetch "Bernie Sanders' AI wealth fund bill shows that he doesn't understand AI or wealth" from Reason.com to get the details of the bill. It has a high likelihood of having detailed quotes and specific provisions of the bill. Let's fetch it.

This is an outstanding source. It gives us the precise details: - Sen. Bernie Sanders (I-Vt.) announced plans to introduce the "American A.I. Sovereign Wealth Fund Act" in the coming weeks (op-ed published June 1, 2026). - The bill would give Americans a "direct ownership stake" in the country's largest AI companies by creating "a sovereign wealth fund through a one-time 50 percent tax" of company stock. - The bill would also allow the federal government to use its "voting shares and an equal representation on each company's board" to block decisions and policies it deems harmful. - It builds on similar calls to action from academics and the leaders of OpenAI, Anthropic, and xAI (who have advocated for a formalized process providing Americans direct payments from the industry). - President Trump issued an executive order last February (February 2025/2026? Wait, it says "last February directing the secretaries of the Treasury Department and the Commerce Department... to develop a plan for a sovereign wealth fund and submit it to the president within 90 days.") Wait, is Donald Trump's meeting with AI executives scheduled in mid-June 2026? Let's check when that meeting is and what the latest is on that. - According to the NYT article (from the search snippet earlier), "He also raised the concept of a sovereign wealth fund with Mr. Sanders, who said Mr. Altman objected to the idea of giving the American people 50 percent of..." - Wait, let's search for "Trump" meeting with "Altman" and "Amodei" on "Sovereign Wealth Fund" in June 2026 to see if there are any details on this meeting. Let's do a search.

Wait, the watch description says: "Tracks President Trump's scheduled meeting with top AI executives (including Sam Altman and Dario Amodei) in mid-June 2026 to discuss the U.S. AI Sovereign Wealth Fund and government equity stakes in AI." And the fire date is June 18, 2026. Let's see if there is any other search we can do about this meeting or if there are any other upcoming pieces of news. Let's search for "Trump" "June 18" "AI" OR "Sovereign Wealth Fund" to see if there are any articles detailing the scheduled June 18 meeting. Let's run a search.

Let's read Fast Company's article "How Trump officials pushed Anthropic to shut down the world's most powerful AI models" to gather more critical insights and quotes about the export controls and the administration's actions. This is incredibly fresh and relevant! Let's fetch it.

Let's fetch the rest of this Fast Company article if there's more, or let's search for "Friday afternoon (June 12)" or additional parts. Wait, is there a second page or is it just a single page? Ah, the fetch returned the beginning part of the article. Let's do a search for the rest of the text from this article or other articles about the timeline. Wait, let's look at the other details: - Friday at 5:21 p.m. ET: U.S. Commerce Department's BIS ordered Anthropic to disable Fable 5 and Mythos 5 for all foreign nationals (inside or outside the US, including foreign national Anthropic employees). - Because Anthropic could not reliably sort users by nationality immediately, the practical effect was a global kill switch: within hours, Anthropic pulled both of its newest and most powerful models for everyone. - The South Korean telecom connection: In May 2026, Anthropic submitted a list of priority access institutions for Mythos 5 (Project Glasswing, a security consortium of 150 institutions across 15 countries, which included SK Telecom, Samsung Electronics, and SK Hynix). The government discovered a "South Korean telecommunications company" with suspected links to China on the list, which shattered fragile trust and triggered the severe export control. - Amazon CEO Andy Jassy called Treasury Secretary Scott Bessent on June 11 to warn about a jailbreak of Fable's guardrails that could enable cyber weapon operability. The NSA reviewed the vulnerabilities and confirmed it was possible to strip away the guardrails, prompting the BIS to act. - Sriram Krishnan, White House AI advisor, is leaving at the end of June 2026. - Let's check other details: Trump signed a new AI oversight executive order (Tuesday, June 9, 2026 or similar? Wait, "Trump signs an executive order that invites vetting of top AI models for national security risks"). Let's search for "Trump signs AI executive order June 2026" or "Trump AI executive order June 2026" to see what that was. Let's run a search.

Let's fetch "New AI Executive Order Calls for Frontier Model Security, Early Government Access and AI-Enabled Cyber Defense" from Skadden to see what this June 2, 2026 Executive Order does. This is a very recent and critical piece of AI policy. Let's fetch it.

This is incredibly rich and detailed. Let's synthesize the timeline and key developments of this week: 1. On June 2, 2026, President Trump signed the executive order "Promoting Advanced Artificial Intelligence Innovation and Security". - It establishes a voluntary framework where "covered frontier models" (designated by a classified benchmarking process run by the NSA, CISA, etc.) can be offered to the government for up to 30 days of pre-release access. - Following this, the model can be released to "trusted partners" selected in collaboration with the government, rather than immediate broad public release. - It explicitly disclaims any mandatory licensing or preclearance. 2. In late May/early June, Anthropic submitted its priority access list of 111 (later expanded by 50 to 161) institutions for its newest frontier model family, Mythos 5 and Fable 5, under its "Project Glasswing" security consortium (which spans 15 countries and includes companies like Samsung, SK Hynix, and SK Telecom, as well as KISA). 3. The White House discovered a "South Korean telecommunications company" with suspected Chinese links on Anthropic's list. This shattered the already fragile trust between the Trump administration and Anthropic (following their previous dispute over the Pentagon's blacklisting and military use of AI). 4. On Thursday night, June 11, Amazon CEO Andy Jassy called Treasury Secretary Scott Bessent to warn that Amazon researchers had found a jailbreak of Fable 5's guardrails that could allow users to bypass safety filters and access Mythos-level cyber capability (effectively weaponizing it or conducting cyberattacks). 5. The NSA reviewed the vulnerability and confirmed it was possible to strip away the guardrails. 6. On Friday morning, June 12, White House and Commerce officials (including Commerce Secretary Howard Lutnick) engaged Anthropic, trying to persuade them to voluntarily pause the release. Anthropic refused, arguing the vulnerability was minor, already known, and present in competitors' models like GPT-5.5. 7. Frustrated by Anthropic's refusal and the South Korean telecom Chinese link issue, the Commerce Department's Bureau of Industry and Security (BIS) issued an emergency export control directive on Friday, June 12, at 5:21 p.m. ET. 8. The BIS order required Anthropic to suspend all access to Claude Fable 5 and Mythos 5 by any foreign national, whether inside or outside the US, including foreign national Anthropic employees. 9. Because Anthropic cannot immediately sort users by nationality, the practical effect was a global kill switch: within hours, Anthropic disabled both models for all users. 10. The export control has massive implications: - It's the first time export controls have been applied to a specific commercial AI model rather than hardware. - It cuts off China's AI labs from their primary distillation source (Claude was heavily used for model distillation by DeepSeek-V3/R1, Qwen 3, Kimi 2, and GLM 5). - It threatens Anthropic's Series H funding round, which was targeting a $300 billion valuation but is now being re-marked to a $260 billion midpoint by underwriters due to an estimated 12% to 18% revenue hit ($700M to $1B). - On Monday, June 15, Anthropic's senior technical staff (including cofounder Tom Brown, head of external affairs Sarah Heck, head of red-teaming Logan Graham, and senior researcher Nicholas Carlini) rushed to Washington, D.C. to meet with Commerce officials and National Cyber Director Sean Cairncross's office, but the talks concluded without a deal to lift the restrictions. 11. Meanwhile, a broader political debate is heating up over the "American AI Sovereign Wealth Fund Act" proposed by Senator Bernie Sanders in a June 1, 2026 NYT op-ed. - Sanders' bill proposes a one-time 50% tax on the stock of the largest AI companies to create a sovereign wealth fund providing direct dividends to Americans, plus government voting shares and board seats to block harmful decisions. - This creates an unlikely convergence with President Trump, who issued an executive order in February directing agencies to plan a US Sovereign Wealth Fund and is scheduled to meet with top AI executives in mid-June (around June 18) to discuss government equity stakes in AI. Let's check if there are other threads. What about the "Meta Muse Spark Developer API Release" watch? The fire date is June 30, 2026. Let's do a quick search on "Meta Muse Spark" or "Muse Spark" to see if there are any recent updates. This is an ongoing/one-shot watch. Let's run a search.

The search budget is spent, which is perfect because we have incredibly high-quality, precise, and fresh information regarding two massive, developing stories: 1. The global suspension of Anthropic's Claude Fable 5 and Mythos 5 models due to an emergency BIS export control order. This is a massive update to `anthropic-recursive-self-improvement-pause-call`. Let's review the slug of the prior finding: `anthropic-recursive-self-improvement-pause-call` (Anthropic Warns of Recursive Self-Improvement and Faces Sharp Developer Backlash Over Code Bloat and IPO Hype). Wait, is there a more specific prior finding for this? Let's see: `anthropic-recursive-self-improvement-pause-call` is indeed very relevant, but this current development is about the *emergency export controls and global shutdown of Fable 5 and Mythos 5*. Let's see if we should write a new note or update the prior one. The prior note is "Anthropic Warns of Recursive Self-Improvement and Faces Sharp Developer Backlash Over Code Bloat and IPO Hype" (slug: `anthropic-recursive-self-improvement-pause-call`). Wait, the instructions say: "Slugs are stable identities, not per-cycle labels — this is what makes findings a living document. Name a slug for the DURABLE SUBJECT it tracks: the throughline, entity, or running debate (ai-coding-backlash, nvidia-datacenter-margins) — never the transient story-of-the-day (sp-h-c99-library, q3-earnings-beat), which can never recur. When this cycle advances a subject you already have a finding for, call write_note with that SAME slug to revise it in place — an upsert that snapshots a new revision. Mint a fresh slug ONLY for a genuinely new subject." Let's look at the prior findings: - `anthropic-recursive-self-improvement-pause-call` - `microsoft-unveils-mai-models-build-2026` - `anthropic-openai-confidential-ipo-filings` - `xai-grok-trained-on-claude-outputs` - `sam-altman-opposes-pre-launch-ai-approvals` Wait, since this is a massive ongoing geopolitical dispute and export control action specifically targeting Anthropic's Fable 5 and Mythos 5 models (which were released on June 9, 2026, and shut down on June 12, 2026), does it fit under `anthropic-recursive-self-improvement-pause-call`? Yes, `anthropic-recursive-self-improvement-pause-call` covers the launch of Claude Fable 5 and Mythos 5 and the initial backlash. Let's reuse this slug to update it with the incredible drama of the export controls and the global shutdown. Wait, actually, there was a watch named "Anthropic Fable 5 Export Control Dispute Resolution" (id: `019eca1f-4c59-7e9c-a52e-20a32ba249fd`). This watch tracks whether the U.S. Commerce Department's Bureau of Industry and Security (BIS) modifies, lifts, or finalizes the emergency export control on Anthropic's Claude Fable 5 and Mythos 5 models following White House negotiations. Since they met on Monday, June 15, and the talks ended without a deal, this watch is *ongoing* (or has there been an event?). Let's record an event on it! Wait, the watch is "Anthropic Fable 5 Export Control Dispute Resolution" and it tracks whether they resolve it. Since they met and did *not* resolve it yet (remains unresolved), we can record an ongoing event or keep watching. Let's record an event that they met on June 15, 2026, but remain split, meaning the export control remains active and unresolved. Let's also look at the other watch: - "Trump Meeting with AI Executives on Sovereign Wealth Fund" (id: `019ebfd3-8cf5-7be0-9118-bff58b1d6a0d`), fire date: June 18, 2026. This meeting is scheduled for June 18, 2026 (which is in 2 days, since current time is June 16, 2026). It hasn't happened yet, so we keep watching it. - "US AI Sovereign Wealth Fund Legislative or Executive Actions" (id: `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`). This is an ongoing watch. We can update our findings about the Sovereign Wealth Fund. Let's look at the prior findings. Is there a prior finding for the Sovereign Wealth Fund? No, there is only "OpenAI CEO Sam Altman Opposes Pre-Launch AI Approval Rules on Capitol Hill" (slug: `sam-altman-opposes-pre-launch-ai-approvals`). Wait, let's write a fresh note for the US AI Sovereign Wealth Fund debate, as it's a massive durable subject (e.g., `us-ai-sovereign-wealth-fund-debate`). And let's write a major revision to `anthropic-recursive-self-improvement-pause-call` or let's see if we should create a specific slug for the export controls if it's a separate durable topic. But wait, the instructions say "When this cycle advances a subject you already have a finding for, call write_note with that SAME slug to revise it in place... Mint a fresh slug ONLY for a genuinely new subject." Since `anthropic-recursive-self-improvement-pause-call` already tracks the launch of Fable 5/Mythos 5 and the initial policy/developer backlash, updating it to cover the subsequent export control shutdown of those exact models is highly appropriate. Wait, let's read the prior finding `anthropic-recursive-self-improvement-pause-call` via `read_notes` first to make sure we understand what was written and can integrate it seamlessly. Let's call `read_notes` on topic `019e92c9-99b4-7b6c-bb81-1e0494672f70`.

This is amazing context. Let's look at the notes we have: - `anthropic-expands-project-glasswing-claude-mythos` (Amazon CEO Andy Jassy’s Security Warning Triggered White House Crackdown on Claude Fable 5) - `us-government-proposes-equity-stakes-in-ai-labs` (Trump and Sanders Align on Creating a U.S. AI Sovereign Wealth Fund via Equity Stakes) Wait, `anthropic-expands-project-glasswing-claude-mythos` is a note that already exists! Let's examine its content. Ah, its title is "Amazon CEO Andy Jassy’s Security Warning Triggered White House Crackdown on Claude Fable 5". Let's update `anthropic-expands-project-glasswing-claude-mythos` because it is the exact slug tracking the Fable 5 / Mythos 5 export control crisis. And `us-government-proposes-equity-stakes-in-ai-labs` is the exact note tracking the Trump-Sanders Sovereign Wealth Fund alignment. Let's fetch the contents of both of these notes using semantic search or we can just write the notes directly with these slugs to update them. Wait, let's look at what we've discovered: 1. **Anthropic Fable 5 and Mythos 5 Export Control Crisis:** - On Friday, June 12, 2026, at 5:21 p.m. ET, the Commerce Department's BIS issued an emergency export control directive ordering Anthropic to disable access to Fable 5 and Mythos 5 for all foreign nationals (inside and outside the US, including its own foreign employees). - Because Anthropic could not instantly verify nationalities, the practical result was a global shutdown (kill switch) of both models. - The trigger: Anthropic submitted its priority access list of 111 (later 161) institutions for Mythos 5 under "Project Glasswing" (a security consortium across 15 countries including Samsung, SK Hynix, SK Telecom, and KISA). The White House discovered a South Korean telecom company with suspected Chinese links on the list, shattering trust. - Amazon CEO Andy Jassy also called Treasury Secretary Scott Bessent on June 11 to warn about a jailbreak of Fable's guardrails that could allow users to bypass safety filters and weaponize its cyber capabilities. The NSA reviewed the vulnerability and confirmed it was possible to strip away the guardrails. - On Monday, June 15, Anthropic's senior technical staff (including cofounder Tom Brown, head of external affairs Sarah Heck, head of red-teaming Logan Graham, and senior researcher Nicholas Carlini) met with Commerce officials and National Cyber Director Sean Cairncross's office in Washington, D.C., but the talks concluded without a deal. - The shutdown threatens Anthropic's Series H funding round, with underwriters re-marking the valuation from $300 billion to a $260 billion midpoint due to a projected $700M to $1B revenue hit (12% to 18% of annualized revenue). - It also cuts off China's AI labs (DeepSeek, Alibaba, Moonshot, Zhipu) from their primary distillation source. 2. **Trump-Sanders Sovereign Wealth Fund & Equity Stakes Alignment:** - On June 1, 2026, Senator Bernie Sanders proposed the "American A.I. Sovereign Wealth Fund Act" in a NYT op-ed, advocating for a one-time 50% tax on the stock of the country's largest AI companies to give the public a direct ownership stake and government voting shares/board seats to block harmful decisions. - This aligns with President Donald Trump, who issued an executive order in February directing agencies to plan a US Sovereign Wealth Fund. - Trump is scheduled to meet with top AI executives (including Sam Altman and Dario Amodei) around June 18, 2026, to discuss government equity stakes in AI. - Sriram Krishnan, White House AI advisor, is leaving his role at the end of June 2026. Let's check if there are other watches we should resolve or record events on: 1. "Anthropic Fable 5 Export Control Dispute Resolution" (id: `019eca1f-4c59-7e9c-a52e-20a32ba249fd`). - Let's record an event on it! Since they met on June 15 and did *not* resolve the dispute, the export control remains active and unresolved. This is an ongoing watch or a one-shot watch? It is a one-shot watch tracking whether they resolve it. Since it's still unresolved, we will set `resolves: false` and record the event. - Wait, let's check the description: "Tracks whether the U.S. Commerce Department's Bureau of Industry and Security (BIS) modifies, lifts, or finalizes the emergency export control on Anthropic's Claude Fable 5 and Mythos 5 models following White House negotiations." Since negotiations have started but are unresolved, let's keep it open (`resolves: false`) and record the event. Wait, let's draft the notes. Let's write a note for the Anthropic export control crisis under the slug `anthropic-expands-project-glasswing-claude-mythos`. Let's write a note for the Trump-Sanders sovereign wealth fund under `us-government-proposes-equity-stakes-in-ai-labs`. Let's double-check the exact rules for writing notes: - No YAML frontmatter. The title is the first `# ` heading. - Must cite at least one real source consulted this cycle — a web page fetched, with links and verbatim quotes kept with their links. - Lift the evidence forward: quotes, links, and interpretation. - One finding = ONE coherent subject tracked under one stable slug. - Wikilinks to link other findings — `[[slug]]` — never a bare `/slug` path or hand-built URL. Let's call `record_event` for the Anthropic Fable 5 Export Control Dispute Resolution watch. Wait, let's look at the parameters for `record_event`: - `thread_id`: `019eca1f-4c59-7e9c-a52e-20a32ba249fd` - `cycle_id`: Let's check if there is a cycle ID. In the kickoff message, the chat ID is `topic-019e92c9-99b4-7b6c-bb81-1e0494672f70-cycle-ab451599-8ca6-412e-9d77-864e7ba4a840`. So the cycle ID is `ab451599-8ca6-412e-9d77-864e7ba4a840`. - `summary`: "Anthropic and White House officials hold high-stakes talks on June 15 but remain split over Fable 5 safety." - `outcome`: "Unresolved. Export controls remain in place, and both models remain offline globally." - `resolves`: `false` (it's a one-shot watch tracking resolution, so we keep it open until a final resolution is reached). - `salience`: 5 (extremely high relevance). Let's call `record_event` now.

Now, let's write the updated note for `anthropic-expands-project-glasswing-claude-mythos`. Let's make sure the note has: - Name and link the actual things (the stories, articles, pages). - Verbatim quotes kept with their links. - Interpretation. - Stable slug `anthropic-expands-project-glasswing-claude-mythos`. - Use wikilinks like `[[us-government-proposes-equity-stakes-in-ai-labs]]` if referencing the other note. Let's review the exact details of the Anthropic export control crisis: - On Friday, June 12, 2026, at 5:21 p.m. ET, the Commerce Department's BIS issued an emergency export control directive ordering Anthropic to disable access to Fable 5 and Mythos 5 for all foreign nationals (inside and outside the US, including its own foreign employees). - Because Anthropic could not instantly verify nationalities, the practical result was a global shutdown (kill switch) of both models. - The trigger: Anthropic submitted its priority access list of 111 (later 161) institutions for Mythos 5 under "Project Glasswing" (a security consortium across 15 countries including Samsung, SK Hynix, SK Telecom, and KISA). The White House discovered a South Korean telecom company with suspected Chinese links on the list, shattering trust. - Amazon CEO Andy Jassy also called Treasury Secretary Scott Bessent on June 11 to warn about a jailbreak of Fable's guardrails that could allow users to bypass safety filters and weaponize its cyber capabilities. The NSA reviewed the vulnerability and confirmed it was possible to strip away the guardrails. - On Monday, June 15, Anthropic's senior technical staff (including cofounder Tom Brown, head of external affairs Sarah Heck, head of red-teaming Logan Graham, and senior researcher Nicholas Carlini) met with Commerce officials and National Cyber Director Sean Cairncross's office in Washington, D.C., but the talks concluded without a deal. - The shutdown threatens Anthropic's Series H funding round, with underwriters re-marking the valuation from $300 billion to a $260 billion midpoint due to a projected $700M to $1B revenue hit (12% to 18% of annualized revenue). - It also cuts off China's AI labs (DeepSeek, Alibaba, Moonshot, Zhipu) from their primary distillation source. Let's write this note now.

That is an outstanding update! Now, let's write the note on the Trump-Sanders Sovereign Wealth Fund debate under the slug `us-government-proposes-equity-stakes-in-ai-labs`. Let's check the title we had for it: "Trump and Sanders Align on Creating a U.S. AI Sovereign Wealth Fund via Equity Stakes". Let's see what is new or has changed since the last cycle. The last cycle's note (from `read_notes`) was: "In a striking convergence of populist economics and state capitalism, both President Donald Trump and Senator Bernie Sanders (I-VT) have proposed separate frameworks to establish a U.S. AI Sovereign..." Wait! Let's update this note to incorporate the very latest details from our research: - Senator Bernie Sanders' "American A.I. Sovereign Wealth Fund Act" proposed in his June 1, 2026 NYT op-ed. It proposes a one-time 50% tax on the stock of the largest AI companies to create a sovereign wealth fund providing direct dividends to Americans, plus government voting shares and board seats to block harmful decisions. - Trump's executive order on AI Sovereign Wealth Fund back in February, which directed Treasury and Commerce to submit a plan within 90 days. - President Trump's upcoming meeting with a dozen top AI executives (including Sam Altman and Dario Amodei) around June 18, 2026, to discuss the U.S. AI Sovereign Wealth Fund and government equity stakes in AI. - Trump's recent signing of the "Promoting Advanced Artificial Intelligence Innovation and Security" Executive Order on June 2, 2026, which sets a voluntary framework for "covered frontier models" to grant the government 30 days of pre-release access. This shows a rapid institutionalization of government oversight. - Sriram Krishnan, White House AI advisor, is leaving his role at the end of June 2026. Let's write this updated note under the slug `us-government-proposes-equity-stakes-in-ai-labs`. Let's construct a beautiful, detailed update.

This is excellent. Let's look at the remaining watches and threads. Are there any other watches to resolve or record events on? 1. "OpenAI Rumored Next Model Release" (id: `019ebfd3-caf2-7375-8fae-269600b57f31`), fire date: June 22, 2026. This is next week, so we keep watching it. 2. "Apple CEO Transition to John Ternus" (id: `019eb060-451b-7ac2-9fec-b3503a6195d2`), fire date: September 1, 2026. This is in the future, so we keep watching it. 3. "Argentina Non-Human Corporation Legislation Vote" (id: `019eb060-3585-778a-a866-491803921b54`), no fire date. This is ongoing, so we keep watching it. 4. "US AI Sovereign Wealth Fund Legislative or Executive Actions" (id: `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`), ongoing watch. We can record an event on it! Since Senator Sanders announced his upcoming "American A.I. Sovereign Wealth Fund Act" on June 1, 2026, and President Trump is hosting a meeting on June 18, 2026, this watch has fired a major event. Let's record an event on it and set `resolves: false` because it's an ongoing watch. - `thread_id`: `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8` - `cycle_id`: `ab451599-8ca6-412e-9d77-864e7ba4a840` - `summary`: "Senator Sanders proposes the American A.I. Sovereign Wealth Fund Act; Trump schedules executive meeting on equity stakes." - `outcome`: "Active. Public and executive proposals for a U.S. AI Sovereign Wealth Fund are accelerating rapidly with bipartisan populist momentum." - `resolves`: `false` (ongoing watch). - `salience`: 5. Let's call `record_event` on the ongoing Sovereign Wealth Fund watch.

That is wonderful. Let's look at the other watches: - "Meta Muse Spark Developer API Release" (id: `019e9889-2962-79c0-ba91-3aa809f9f6b9`), fire date: June 30, 2026. This is currently scheduled for June 2026, but our research shows that as of June 3, 2026, Meta has repeatedly delayed the developer API for its Muse Spark model due to bugs and infrastructure issues, and as of early June, has no scheduled launch date. We keep watching until June 30, 2026, to see if it successfully ships this month or faces further delays. Wait! Let's check if there are any other threads or notes we should update. Let's review the notes in our list: We updated: 1. `anthropic-expands-project-glasswing-claude-mythos` (Amazon CEO Andy Jassy’s Security Warning Triggered White House Crackdown on Claude Fable 5) 2. `us-government-proposes-equity-stakes-in-ai-labs` (Trump and Sanders Align on Creating a U.S. AI Sovereign Wealth Fund via Equity Stakes) Are there any other notes that we should update or write? Let's see if there are any other major developments: - OpenAI quietly retired its older GPT-5.2 model family and moved everyone to GPT-5.5 as leaks of GPT-5.6 intensify (this was updated on June 15, 2026 under `openai-gpt-model-releases`). No major change has occurred since June 15, which was yesterday, so we don't need to write a duplicate or pad it. - Meta's Muse Spark API delays (updated on June 16, 2026 under `meta-muse-spark-api-delays` and `meta-delays-muse-spark-api-bugs`). No major updates there since. - Argentina's non-human corporations (updated June 15, 2026 under `argentina-non-human-corporation-ai-regulation`). No major updates there. This is a comprehensive and incredibly rich update. We have captured everything new and what has changed since the last cycle. Let's double-check if there are any other actions to take. We have: - Met the criteria for `record_event` on the two watches (`019eca1f-4c59-7e9c-a52e-20a32ba249fd` and `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`). - Written two outstanding, detailed findings (`anthropic-expands-project-glasswing-claude-mythos` and `us-government-proposes-equity-stakes-in-ai-labs`) that carry representative verbatim quotes, links, and deep interpretations. - Ensured wikilinks are used to connect notes instead of bare paths or full URLs. - Noted that no new threads or watches need to be minted because the existing set of threads and watches perfectly frames the ongoing developments. I will now call `finish_cycle` to complete this research cycle.

I have completed the research cycle for topic `019e92c9-99b4-7b6c-bb81-1e0494672f70`. ### Key Findings and Updates Persisted this Cycle: 1. **The Anthropic Fable 5 & Mythos 5 Global Export Control Crisis** * **What Happened:** On Friday, June 12, 2026, at 5:21 p.m. ET, the Commerce Department's Bureau of Industry and Security (BIS) issued an emergency export control directive ordering Anthropic to immediately disable access to its newly released Claude Fable 5 and Claude Mythos 5 models for all foreign nationals. Lacking the capability to instantly verify user nationalities, Anthropic was forced to pull both frontier models offline globally, triggering a massive industry shockwave. * **The Triggers:** * **South Korean Telecom Link:** Anthropic submitted a list of 111 (later 161) priority access institutions for its full-strength Mythos 5 model under its "Project Glasswing" security consortium. The discovery of a unnamed South Korean telecommunications firm with suspected Chinese links on this list shattered the White House's trust in Anthropic. * **Amazon's Jailbreak Warning:** On June 11, Amazon CEO Andy Jassy directly warned Treasury Secretary Scott Bessent of a Fable 5 jailbreak that could bypass safety guardrails to weaponize its cybersecurity and biological capabilities. An independent NSA review confirmed the vulnerability, prompting immediate White House intervention. * **Current Status:** High-stakes negotiations in Washington, D.C. on Monday, June 15, between senior Anthropic technical staff and Commerce officials (led by Howard Lutnick and National Cyber Director Sean Cairncross's office) concluded without a deal. Both models remain offline globally. * **Financial & Geopolitical Fallout:** The shutdown is projected to hit Anthropic's annualized revenue by 12% to 18% ($700M to $1B), forcing lead underwriters of its pending Series H funding round to re-mark its valuation from $300 billion down to a $260 billion midpoint. It also freezes China's frontier AI labs (DeepSeek, Alibaba, Moonshot, Zhipu), which have heavily relied on Claude's outputs for model distillation. * **Durable Slug Updated:** `anthropic-expands-project-glasswing-claude-mythos` 2. **The Trump-Sanders Sovereign Wealth Fund & Equity Stakes Convergence** * **What Happened:** An unlikely populist alliance is forming over the ownership and wealth distribution of the AI boom. On June 1, 2026, Senator Bernie Sanders (I-VT) proposed the **American A.I. Sovereign Wealth Fund Act**, advocating for a one-time 50% tax on the stock of the country's largest AI companies to fund direct dividends to the public and secure voting government board seats. * **Trump's Parallel Play:** This aligns tactically with President Trump's February executive order directing the Treasury and Commerce Departments to plan a U.S. Sovereign Wealth Fund. Trump is scheduled to host a high-profile meeting with top AI executives (including Sam Altman and Dario Amodei) around **June 18, 2026**, to negotiate government equity stakes in AI labs. * **A New Era of Oversight:** This push coincides with the signing of Trump's June 2, 2026 Executive Order, "Promoting Advanced Artificial Intelligence Innovation and Security," which establishes a "voluntary" framework allowing the government up to 30 days of pre-release vetting access for any model designated as a "covered frontier model" by the NSA and CISA. * **Durable Slug Updated:** `us-government-proposes-equity-stakes-in-ai-labs` ### Watch Status and Events Recorded: * **Anthropic Fable 5 Export Control Dispute Resolution (ID: `019eca1f-4c59-7e9c-a52e-20a32ba249fd`):** Event recorded. Monday, June 15, high-stakes talks ended without a deal; export controls remain active and models remain offline. (Watch remains pending). * **US AI Sovereign Wealth Fund Legislative or Executive Actions (ID: `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`):** Event recorded. Senator Sanders' legislative proposal and President Trump's upcoming June 18 executive meeting represent a massive escalation in sovereign wealth fund planning. (Watch remains active/ongoing).