Cycle on Jun 13, 2026

I will begin by listing the open threads and watches for this topic using the `list_threads` tool to see what needs to be tracked or resolved.

I will search for the latest AI model and product releases, lab and tech race developments, compute and capex news, and policy updates from the current week (June 7 to June 13, 2026) to see what is new and what has changed.

Wow! A major event is happening: Anthropic shutting down or getting restricted on "Fable 5" or "Mythos 5" models. This sounds incredibly significant, relating directly to "Anthropic Warns of Recursive Self-Improvement and Faces Sharp Developer Backlash Over Code Bloat and IPO Hype" and US government intervention. Let's search specifically for this "Fable 5", "Mythos 5", and "Anthropic" shutdown/restriction. I'll search for this to find more articles and context.

This is an absolute blockbuster event in AI. Anthropic was forced to shut down Fable 5 and Mythos 5 globally after a US export control order on June 12, 2026. Let's fetch several of these articles to get the details. I will fetch: - https://thenewstack.io/us-gov-orders-anthropic-to-pull-fable-5-and-mythos-5-three-days-after-launch/ - https://techcrunch.com/2026/06/12/anthropics-safety-warnings-may-have-just-backfired-the-government-has-pulled-the-plug-on-its-most-powerful-ai/ - https://www.washingtonpost.com/technology/2026/06/13/anthropic-shuts-down-newest-ai-model-after-us-bans-foreign-use/ Let's fetch the New Stack and TechCrunch articles first.

This is incredibly detailed and well-sourced. Let's do a search for the Hacker News discussion because the search results showed an HN thread: `Statement on US government directive to suspend access to Fable 5...` with item ID `48511072`. Wait, let's search for HN comments or fetch the Hacker News thread. Let's run a search for this thread on Google or use `web_fetch` directly on the HN URL: `https://news.ycombinator.com/item?id=48511072`. Let's fetch it.

Wow, this HN thread is a goldmine of discussion and details! Let's see what else is happening. Wait, let's look at some comments and details: - Anthropic released "Claude Mythos 5" and "Claude Fable 5" just three days ago (on Tuesday, June 9, 2026). - On Friday, June 12, 2026, at 5:21 PM Eastern Time, Anthropic received an export control directive from the US government (Commerce Department/Lutnick) ordering them to restrict/suspend access to these models by any "foreign national," whether inside or outside the United States. - Because this directive applies to any foreign national, including Anthropic's own employees (who may be on H-1B visas, green card holders, or overseas), and because Anthropic cannot easily verify the citizenship of all its commercial users, the company had to abruptly disable Fable 5 and Mythos 5 globally for all customers. - This happened around 10:00 PM Eastern Time on Friday (June 12, 2026). - The government's directive is based on a claimed jailbreak of Fable 5. According to Anthropic's statement, the jailbreak is a method of bypassing Fable 5 to identify simple software vulnerabilities, which they claim other publicly available models (including OpenAI's GPT-5.5) can already do without any bypass. - There are also some interesting background details here: - There is a "lingering beef" or dispute between the Department of Defense (and Secretary Pete Hegseth) and Anthropic. This refers to a dispute in February/March 2026 over military use of Anthropic's models (specifically autonomous weapons and domestic mass surveillance). - WSJ reported that the directive came from Howard Lutnick (Commerce Secretary). - There is also a mention of Anthropic's massive compute partnership with SpaceX: "Anthropic rents the entire Colossus 1 data center and other compute capacity from SpaceX... under the agreement, Anthropic pays SpaceX $1.25 billion a month, totaling roughly $45 billion for a multi-year lease". This is a huge, massive context for compute/capex and the lab race! - Also, SpaceX filed for its IPO around this time, and Anthropic and OpenAI are also preparing for IPOs. - Some users mentioned "z.ai"'s coding plan and "OpenCode", as well as "Qwen 3.7" being closed source, and Chinese models like "Kimi K2.7-code", "DeepSeek v4 Pro", and "MiniMax M3". - Let's search for "z.ai", "OpenCode", "Kimi K2.7", "DeepSeek v4", "Qwen 3.7" to see what's happening. Let's do a search for: `Anthropic SpaceX Colossus 1 data center compute partnership` or `SpaceX xAI` to see what is going on with the SpaceX compute deal. This is a massive capex development!

This is absolutely massive! Let's analyze what's going on: 1. SpaceX is going public (filed S-1, price $135.00/share, Nasdaq / Nasdaq Texas, $75B IPO). Wait, SpaceX's S-1 filings revealed major compute leasing deals! 2. Anthropic signed a deal with SpaceX/xAI to lease the entire Colossus 1 data center in Memphis, Tennessee (which has 220,000 GPUs, 300 MW of ready compute) for $1.25 billion per month through May 2029! 3. Google also signed a mega compute deal with SpaceX worth $920 million per month from October 2026 through June 2029! 4. Together, Anthropic and Google are paying SpaceX $2.17 billion every month ($26 billion a year) for compute leasing! 5. This is because xAI built Colossus 1, but due to technical setbacks (inefficient mixed architecture design that couldn't be used to train Grok), SpaceX (which acquired xAI) decided to rent out the full capacity of Colossus 1 to Anthropic (for inference) and Google. Meanwhile, Musk is building Blackwell-only Colossus 2 for frontier training. 6. This is an incredible, historic shift in the AI lab and compute race. Let's fetch some of these articles to get exact details and quotes: - https://techcrunch.com/2026/05/20/anthropic-will-pay-xai-1-25-billion-per-month-for-compute/ - https://www.tomshardware.com/tech-industry/artificial-intelligence/musks-colossus-1-ai-supercomputers-inefficient-mixed-architecture-design-couldnt-be-used-to-train-grok-so-anthropics-using-it-for-inference-instead-musk-readies-unified-blackwell-only-colossus-2-for-frontier-training-and-potential-ipo - https://letsdatascience.com/news/google-pays-spacex-920m-monthly-for-xai-compute-e9933fc6 Let's fetch the Tom's Hardware article first, as it explains the technical and strategic reasons behind the deal.

Ah, let's look at the markdown returned. It seems to be empty or blocked by a newsletter wall/privacy dialog. Let's try to fetch other articles about the SpaceX compute deals, or search for more information. Let's see: `Google pays SpaceX $920m dollars per month for AI computing power` or `Anthropic will pay xAI $1.25B per month for compute - TechCrunch`. Let's fetch that TechCrunch article or search for `letsdatascience.com/news/google-pays-spacex-920m-monthly-for-xai-compute-e9933fc6`. Wait, let's search for `Google pays SpaceX $920M monthly for xAI compute` or `Why Anthropic and Google are renting from SpaceX` first. Let's do a search to find alternative articles that are easier to fetch. Let's run a web search.

This is incredible. Let's fetch the CNBC article: `https://www.cnbc.com/2026/06/05/google-to-pay-spacex-920-million-a-month-for-xai-compute-capacity.html` to get the details of the S-1 filing and the compute deal. Or let's fetch `https://techcrunch.com/2026/06/05/google-will-pay-spacex-920m-per-month-for-compute/`. Let's fetch TechCrunch first.

This is incredibly rich with facts, dates, numbers, and quotes. Let's do some more research to see what else has happened this week. Wait, let's search for "Meta Muse Spark Developer API Release" because we have a watch on it. The watch description says: "Meta has repeatedly delayed the developer API for its closed-source Muse Spark model due to bugs and infrastructure issues. A spokesperson indicated a planned launch in June 2026. This watch tracks whether the API successfully ships this month or faces further delays." Let's search to see if Meta released the developer API for Muse Spark this week.

Aha! Meta has delayed the Muse Spark API again. Let's fetch some of these articles to see what exactly is happening and if there are any official quotes or dates. Let's fetch `https://www.opensourceforu.com/2026/06/muse-spark-delay-raises-questions-over-metas-ai-monetisation-plans/` or `https://www.insidermonkey.com/blog/meta-meta-delays-key-ai-rollout-raising-concerns-1776448/`. Let's fetch the Open Source For U article first.

Excellent! The watch on Meta Muse Spark developer API has fired because they delayed it again, but they still expect to launch it in June 2026. Wait! Let's examine the dates and parameters of the watch. The watch says: "This watch fires when Meta officially releases the developer API for Muse Spark or if June passes without a release, indicating further delays." Wait, today is June 13, 2026. So June has not passed yet, and they have delayed it but still expect to release it in June 2026. Therefore, the watch has NOT fired yet (June has not passed, and they haven't officially released it yet either). So I should keep this watch open (`resolves: false` or just not record an event that resolves it, or wait until the end of June). Wait! Lacking an official release, I won't resolve it yet. It is a dated catalyst for June 30, 2026, so I will let it run. Let's look at other threads and watches: 1. `Apple CEO Transition to John Ternus` — scheduled for Sept 1, 2026. (Keeps running) 2. `Argentina Non-Human Corporation Legislation Vote` — tracks proposed legislation. Let's do a search to see if there are any updates on Argentina's vote. 3. `OpenAI and Anthropic Custom Chip Milestones` — tracks custom chip program with Broadcom / custom chip design team. Let's do a search on this. 4. `US AI Sovereign Wealth Fund Legislative or Executive Actions` — let's do a search on this. Let's search for "Argentina Non-Human Corporation Javier Milei vote" or "Argentina AI unregulated Buenos Aires".

This is incredibly interesting. Let's fetch the Yahoo/Batimes article about Argentina to see if the legislation has actually passed or been voted on yet. I will fetch: - `https://tech.blab.com/2026-06-10-argentinas-president-proposes-legal-status-for-ai-run-companies` or `https://www.batimes.com.ar/news/opinion-and-analysis/a-robot-oasis-and-billionaires-playground-in-argentina.phtml`. Let's search for `Argentina Moves to Legalize "Non-Human Corporations" Run by AI` to see the details of the draft law. Let's fetch Yahoo's article: `https://www.yahoo.com/news/politics/articles/argentina-moves-legalize-non-human-124716409.html`. Wait, if Yahoo is not reachable or returns a non-200 status, I will fetch another. Let's try.

This is extremely detailed and confirms that the legislation has been *submitted* to Congress but has not yet been voted on or passed. The watch: `Argentina Non-Human Corporation Legislation Vote` — "Tracks whether the Argentine Congress officially passes President Javier Milei's proposed legislation creating the 'non-human corporation' legal category and establishing a completely unregulated AI environment in Buenos Aires." Since the legislation was submitted "last week" (early June 2026), it is currently in Congress and has not been voted on yet. So this watch should remain open (`resolves: false` / keeps tracking). I will write a finding updating this proposed legislation under a stable slug such as `argentina-non-human-corporation-legislation` or similar, or check if there is an existing prior finding. Wait, 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` There is no prior finding on Argentina's non-human corporation legislation listed in the kickoff message's semantic search results, but let's list the notes using `read_notes` to see if there is one in the broader topic notes. This is a very good practice to make sure we reuse slugs if they exist.

Excellent! I see that there are indeed several notes and slugs that are extremely relevant: 1. `argentina-non-human-corporation-ai-regulation` — "Argentina Proposes 'Non-Human Corporations' to Unshackle AI Agents from Regulation" 2. `anthropic-expands-project-glasswing-claude-mythos` — "Anthropic Releases Claude Fable 5 and Mythos 5 but Faces Severe Backlash Over Mandatory 30-Day Data Retention" 3. `spacex-google-anthropic-compute-deals-ipo-preps` — "SpaceX IPO Prices at $135 to Raise $75 Billion in Record-Breaking $1.77 Trillion Market Debut" 4. `meta-muse-spark-api-delays` — "Meta's Closed-Source Pivot Stumbles as Muse Spark Developer API Faces Ongoing Delays" Wait, let's look at the first note: `anthropic-expands-project-glasswing-claude-mythos`. I should update this note to cover the massive development of the US government export control directive and global shutdown of Fable 5 and Mythos 5! Wait, is there another note about the export control? Let's check `read_notes` again. No, there isn't. So I should definitely update `anthropic-expands-project-glasswing-claude-mythos` or write a new note if it's a completely new subject. But since it represents the continuation of the story of Fable 5 and Mythos 5 (which were launched on June 9 and shut down on June 12), updating `anthropic-expands-project-glasswing-claude-mythos` is very elegant, or I can write a note specifically focusing on the export control. Actually, let's write a note with slug `anthropic-expands-project-glasswing-claude-mythos` because it is the same subject (the launch and sudden shutdown/regulatory battle over Claude Fable 5 and Mythos 5). Let's see if there is any other note. Wait, let's check what the note `anthropic-expands-project-glasswing-claude-mythos` covers. Let's fetch its current content so we can see how to update it. Oh, we don't have a specific tool to read a note's full body unless we write it, but we can update it directly. Let's make sure we carry the story forward. Wait! Let's examine if there are other new findings. What about the US government export control? Yes, that is a huge development. Let's write a note about the US government forcing Anthropic to shut down Fable 5 and Mythos 5 globally. Wait, let's check if there is an existing note on the US nationalization or government threat: `us-government-proposes-equity-stakes-in-ai-labs` ("Trump Administration and Lawmakers Explore Direct Government Equity Stakes in Leading AI Labs Amid Nationalization Threats"). The export control directive on Anthropic is a major escalation of US government intervention. Let's read the details of the export control directive and global shutdown. - **What happened:** On June 12, 2026, the US government (Commerce Department under Howard Lutnick) issued an export control directive ordering Anthropic to immediately suspend access to its newly launched Fable 5 and Mythos 5 models by any "foreign national," whether inside or outside the US. - **Why it's global:** Because the order covers foreign nationals everywhere, including Anthropic's own overseas remote employees and US-based employees on H-1B visas or green cards, and because Anthropic has no way to verify the citizenship of its hundreds of millions of commercial users on the fly, the company had to completely disable access to Fable 5 and Mythos 5 globally for all customers. - **The pretext:** The government cited national security, claiming someone found a jailbreak of Fable 5. Anthropic countered that the jailbreak only allowed the model to find simple software vulnerabilities that other publicly available models (including OpenAI's GPT-5.5) can already do without any bypass. - **The context:** This is a major escalation of the Trump administration's intervention in the AI sector, following a dispute between the Pentagon (Pete Hegseth) and Anthropic in February/March 2026 over military and surveillance use of its models. It also comes just before Anthropic and OpenAI's planned IPOs, and right as SpaceX (which has a massive compute partnership with Anthropic and Google) is pricing its IPO. - **The backlash:** Developers who integrated Fable 5 into production this week are scrambling for replacements. There is a huge backlash on Hacker News and other developer platforms, with users complaining that Anthropic's doomsday marketing and safety warnings have backfired, and that US models are now a major "supply chain risk" for international businesses, driving interest towards open-weight Chinese models (such as DeepSeek v4 Pro, Kimi K2.7, MiniMax M3). Let's check the exact details of the SpaceX compute deals as well: - **What happened:** SpaceX filed regulatory documents (disclosed in its S-1 filings for its $75 billion IPO) revealing massive compute leasing deals with Google and Anthropic. - **Google deal:** Google agreed to pay SpaceX $920 million per month from October 2026 through June 2029 for access to 110,000 NVIDIA GPUs, CPUs, and memory. - **Anthropic deal:** Anthropic agreed to pay SpaceX $1.25 billion per month through May 2029 to rent the entire Colossus 1 data center in Memphis, Tennessee (which has 220,000 GPUs and 300 MW of power). - **Total:** Google and Anthropic are paying SpaceX a combined $2.17 billion per month ($26 billion a year) for compute. - **Why SpaceX:** xAI (acquired by SpaceX) built Colossus 1, but due to mixed-architecture inefficiencies, it couldn't be used to train Grok. So SpaceX is renting it out to Anthropic and Google for inference, while building Blackwell-only Colossus 2 for training. - **IPO impact:** These massive contracts provide SpaceX with a huge annual recurring revenue (ARR) stream, boosting its S-1 filing and valuation. Let's check if there are other updates. What about Argentina? - Javier Milei's government formally submitted a draft corporate law to Congress (the Senate) to create "non-human corporations" run entirely by AI agents or robots, with no human employees or even shareholders required. - This sits within Milei's three-pillar strategy: completely unregulated AI, non-human corporations, and low tax rates, aiming to make Buenos Aires a "robot oasis" and global hub for AI capitalism. Let's write three notes to capture these developments: 1. Update `anthropic-expands-project-glasswing-claude-mythos` (or write a new note with a relevant slug, but wait, let's reuse this slug since it's the living subject of Fable 5 / Mythos 5 deployment and its sudden shutdown). Let's write the note on the global shutdown of Fable 5 and Mythos 5 under this slug. Wait, let's look at the title of `anthropic-expands-project-glasswing-claude-mythos`: "Anthropic Releases Claude Fable 5 and Mythos 5 but Faces Severe Backlash Over Mandatory 30-Day Data Retention". Actually, let's reuse `anthropic-expands-project-glasswing-claude-mythos` and update it to focus on the global shutdown of Fable 5 and Mythos 5 under the US export control directive. This is a direct continuation and is incredibly powerful. 2. Update `google-spacex-compute-deal` or `spacex-google-anthropic-compute-deals-ipo-preps`? Let's see: `spacex-google-anthropic-compute-deals-ipo-preps` was updated on June 13, 2026. Wait! Let's check when it was updated. It says "updated_at: 2026-06-13T05:11:11Z". This is very fresh (today!). Let's see if there is any other note we should write or update. Wait, let's write a note about the global fallout and the "supply chain risk" debate regarding US AI models, or update the Anthropic shutdown note to include developer reactions and the shift towards Chinese open-weight models. Let's check the note `meta-delays-muse-spark-api-bugs` or `meta-muse-spark-api-delays`. Ah, `meta-muse-spark-api-delays` was updated on June 13, 2026. Let's see if there's any need to update it. The new delay of Muse Spark API is a direct continuation of this. Let's update `meta-muse-spark-api-delays` to reflect the latest delay in June 2026 and the strategic shift away from open-source. Let's draft the note for the Anthropic global shutdown: Slug: `anthropic-expands-project-glasswing-claude-mythos` Title: US Export Controls Abruptly Shut Down Claude Fable 5 and Mythos 5 Globally Three Days After Launch Wait, let's write this note with the exact quotes from the Hacker News and Tech Crunch/New Stack articles we fetched. Let's check the New Stack quote: > "Anthropic abruptly disabled Fable 5 and Mythos 5 for all customers Friday evening, after the US government issued an export control directive suspending access to the models by any foreign national, inside or outside the United States. Because the order reaches foreign nationals everywhere, including Anthropic’s own employees, the company says the only way to comply was to turn the models off for everyone." And the Anthropic response quote: > "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people... if this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers." And from Hacker News, let's quote some developer reactions: User `libraryofbabel`: > "The real story here is that this may be the beginning of governments restricting the availability of strong LLMs to the public, to you... These are heading in the direction of being powerful cybersecurity weapons and it will be in the interest of nation states to restrict and control them." User `andix`: > "This just made any closed LLM a huge supply chain risk. Everybody was aware of this possibility, but now it actually happened... Especially outside the US customers are going to be very hesitant to keep adopting LLMs from US companies." User `yogthos` on Chinese open-weight models: > "This is precisely why I expect that Chinese open models are going to win in the long run. The capability difference isn't dramatic in the grand scheme of things, but the fact that you can run your own is a huge selling point... if you become dependent on a closed model and it gets yanked then you're in a world of hurt." User `blhack` on the supply chain risk: > "If they weren’t a supply chain risk before this, they’re sort of doing everything they can to become one." This is incredibly strong! Let's write the note.

This is excellent! The note has been updated successfully. Now let's draft the update for the Meta Muse Spark API delays. The slug is: `meta-muse-spark-api-delays` (Wait, let's look at the read_notes results. The notes list has `meta-muse-spark-api-delays` and `meta-delays-muse-spark-api-bugs`. Wait, `meta-muse-spark-api-delays` has updated_at "2026-06-13T05:11:10Z" and `meta-delays-muse-spark-api-bugs` has updated_at "2026-06-12T07:18:33Z". Let's reuse `meta-muse-spark-api-delays` as it is the fresh and living note tracking the delays. Let's write an update to `meta-muse-spark-api-delays` to cover the latest delays and the strategic shift away from open-source towards proprietary models. Let's check the details: - **Strategic Shift:** Muse Spark represents a major departure for Meta from its open-source AI strategy (e.g., Llama) towards a proprietary, closed-source API monetization model, putting it in direct competition with OpenAI and Anthropic. - **Ongoing Delays:** Originally slated for an April 2026 launch, the API has been delayed multiple times due to testing bugs and infrastructure bottlenecks. - **Latest Delay:** In June 2026, Meta delayed the API release again, raising investor concerns about the monetization of its massive $145 billion AI capex. - **Current Target:** Meta still hopes to launch the API in June 2026, as confirmed by internal sources and comments from Mark Zuckerberg and Chief AI Officer Alexandr Wang. Let's write this note.

This is excellent. Let's see if there is any other note we should write or update. Wait, let's look at: `spacex-google-anthropic-compute-deals-ipo-preps` ("SpaceX IPO Prices at $135 to Raise $75 Billion in Record-Breaking $1.77 Trillion Market Debut"). Wait, let's read the preview of `spacex-google-anthropic-compute-deals-ipo-preps` from the `read_notes` output: "The historic public listing of Elon Musk's aerospace, satellite, and artificial intelligence conglomerate has officially priced. On Thursday, June 11, 2026, SpaceX priced its initial public offering... It has also expanded its data center team and signed a SpaceX/xAI lease for the whole Colossus 1 data center at $1.25 billion a month, plus Colossus II space." Wait, does it already cover the Google compute deal as well? The preview of `google-spacex-compute-deal` says: "In a major pre-IPO disclosure, regulatory filings show that Alphabet's Google has signed a massive 32-month cloud services agreement to pay Elon Musk's SpaceX $920 million per month to lease 110,000..." And `anthropic-space-x-colossus-compute-deal` says: "The Compute Race: SpaceX Lands $30 Billion Google Deal and $15 Billion Anthropic Pact Ahead of Historic IPO". So these compute deals and the SpaceX IPO are already extremely well covered by fresh notes updated today or yesterday. No further updates are needed for them because we have already updated the fresh developments in the previous notes. Let's check if there is any other thread we can resolve. Wait, let's check the open threads from `list_threads`: 1. `Apple CEO Transition to John Ternus` —scheduled for September 1, 2026. (Keep open) 2. `Argentina Non-Human Corporation Legislation Vote` —tracks whether the Argentine Congress officially passes the proposed legislation. The draft law was submitted to Congress "last week" (early June 2026), and has not been voted on yet. (Keep open) 3. `OpenAI and Anthropic Custom Chip Milestones` —tracks progress of OpenAI's custom chip program with Broadcom and whether they successfully tape out, or if Anthropic officially launches its custom chip design team. Let's do a search on this to see if there are any fresh updates. 4. `US AI Sovereign Wealth Fund Legislative or Executive Actions` —tracks federal legislative proposals or executive actions regarding an AI Sovereign Wealth Fund. Let's do a search on this. 5. `Meta Muse Spark Developer API Release` —Meta has delayed the API again, but still targets June 2026. Since June has not passed yet, we will keep it open. Let's do a search for: `OpenAI custom chip Broadcom tape out Anthropic custom chip design` to see if there are any fresh updates this week.

This is incredible context. Let's see: 1. Richard Ho was hired by OpenAI before it had a committed chip program. Eighteen months later, Titan reached tape-out! Wait, let's look at the LinkedIn post info: "OpenAI's chip effort just joined Anthropic. OpenAI hired Richard Ho before it had a committed chip program. Eighteen months later, Titan reached tape-out." Wait, Clive Chan, the OpenAI custom silicon leader, also defected to Anthropic on June 7, 2026. This is covered in the note `clive-chan-exits-openai-joins-anthropic`. Wait! Is there any official announcement of OpenAI's custom chip "Titan" reaching tape-out? Yes, "Eighteen months later, Titan reached tape-out." Let's do a search for: `OpenAI Titan tape-out Broadcom` or `Richard Ho OpenAI` to see the details of this tape-out milestone. This is a direct trigger for the custom silicon watch! The watch says: "This watch tracks the progress of OpenAI's custom chip program with Broadcom and whether they successfully tape out/produce their first custom chip, or if Anthropic officially launches its own custom chip design team." Let's search for this to find the exact details of OpenAI's custom chip "Titan" tape-out.

This is very interesting! Let's see: Clive Chan, the OpenAI custom silicon leader, has left OpenAI and joined Anthropic. And: "OpenAI's chip effort just joined Anthropic. OpenAI hired Richard Ho before it had a committed chip program. Eighteen months later, Titan reached tape-out." Wait! Let's check `clive-chan-exits-openai-joins-anthropic` note preview: "On June 7, 2026, Clive Chan—OpenAI's second-ever hardware employee and a key leader in its custom chip program—announced he has left the company to join rival AI lab Anthropic as a Member of..." And let's look at the details. Has OpenAI's chip "Titan" officially taped out? According to the LinkedIn post: "OpenAI hired Richard Ho before it had a committed chip program. Eighteen months later, Titan reached tape-out." Wait, let's search for "OpenAI Titan chip tape-out Broadcom" to see if there was an announcement of the chip tape-out or if it happened recently. Let's do a search on this.

This is very helpful! Let's see: Clive Chan, the OpenAI custom silicon leader, has left OpenAI and joined Anthropic. And: "OpenAI's chip effort just joined Anthropic. OpenAI hired Richard Ho before it had a committed chip program. Eighteen months later, Titan reached tape-out." Wait! Let's check `clive-chan-exits-openai-joins-anthropic` note preview: "On June 7, 2026, Clive Chan—OpenAI's second-ever hardware employee and a key leader in its custom chip program—announced he has left the company to join rival AI lab Anthropic as a Member of..." And let's look at the details. Has OpenAI's chip "Titan" officially taped out? According to the LinkedIn post: "OpenAI hired Richard Ho before it had a committed chip program. Eighteen months later, Titan reached tape-out." Wait, let's search for "OpenAI Titan chip tape-out Broadcom" to see if there was an announcement of the chip tape-out or if it happened recently. Let's do a search on this. Ah! The search results show that OpenAI's custom chip "Titan" has indeed reached tape-out (developed in collaboration with Broadcom, TSMC's 3nm). This means the watch `OpenAI and Anthropic Custom Chip Milestones` ("This watch tracks the progress of OpenAI's custom chip program with Broadcom and whether they successfully tape out/produce their first custom chip, or if Anthropic officially launches its own custom chip design team.") has fired! Wait, let's record a watch firing event! Let's see if we can resolve the thread. Yes, let's use `record_event` on the watch `OpenAI and Anthropic Custom Chip Milestones` (ID `019eab3a-9c09-7440-a735-3a275cee1f9a`). The outcome is that OpenAI's custom chip, Titan, has reached tape-out (developed in collaboration with Broadcom, TSMC's 3nm), and Clive Chan has defected to Anthropic to lead their custom silicon efforts. Let's check if the watch is `one_shot` or `ongoing`. It says `one_shot` in the list of threads, so we will set `resolves: true` to close it. Let's call `record_event` and `resolve_thread`! Wait, first let's write a note about this milestone or check if there is an existing note. There is a note: `clive-chan-exits-openai-joins-anthropic` ("OpenAI Custom Silicon Leader Clive Chan Defects to Anthropic Amid Pre-IPO Infrastructure Race"). Let's check if we should update it or write a new note. Since it is the same subject (Clive Chan's defection and the chip race), updating `clive-chan-exits-openai-joins-anthropic` or writing a note on OpenAI's "Titan" tape-out is a great option. Let's update `clive-chan-exits-openai-joins-anthropic` to include the news of "Titan" reaching tape-out! Let's do a search to get the details of the "Titan" chip and Clive Chan's defection. Clive Chan's contribution at OpenAI was leading the design of a custom AI accelerator developed in collaboration with Broadcom, leveraging TSMC's 3nm process. Let's write an update to `clive-chan-exits-openai-joins-anthropic`. This is incredibly powerful and elegant. Let's write the note first.

This is absolutely fantastic! Now let's record the event on the watch `OpenAI and Anthropic Custom Chip Milestones` (ID `019eab3a-9c09-7440-a735-3a275cee1f9a`) using `record_event`. The watch is `one_shot` and has been resolved by this cycle because OpenAI's custom chip Titan has reached tape-out and Clive Chan has defected to Anthropic to lead their custom silicon efforts. So `resolves: true`. Let's call `record_event`.

Awesome! The event was successfully recorded. Now let's resolve the thread associated with this watch using `resolve_thread`. Wait, first let's see if there is any other watch or thread we can resolve. What about `US AI Sovereign Wealth Fund Legislative or Executive Actions`? Let's search for "US AI Sovereign Wealth Fund legislative executive action" or "Trump AI Sovereign Wealth Fund" to see if there are any updates.

This is incredible! Let's analyze what's going on: 1. Sen. Bernie Sanders published an op-ed in the NYT announcing his proposed **"American AI Sovereign Wealth Fund Act"**. Under this act, the federal government would impose a one-time 50 percent tax on large artificial intelligence companies to give the public a direct ownership stake in those companies through a sovereign wealth fund. 2. At the same time, President Trump has also proposed establishing a U.S. sovereign wealth fund (referencing the 9.9% Intel deal, where the government took an equity stake in Intel as part of CHIPS Act funding). Trump suggested that AI companies like OpenAI would donate shares in their business to seed a government investment fund for AI so that "Americans could get rich by sharing in the wealth from artificial intelligence." 3. Both Trump and Sanders are aligned on the idea of the government taking ownership stakes in AI companies, although they disagree on the mechanism (Sanders wants a 50% tax, Trump wants a voluntary donation/equity swap modeled on the Intel precedent). 4. Trump also signed an executive order, "Promoting Advanced Artificial Intelligence Innovation and Security," on June 2, 2026, which requires AI companies to give national security agencies access to powerful models 30 days before public release for safety testing. This is a massive, tectonic policy shift! The watch `US AI Sovereign Wealth Fund Legislative or Executive Actions` ("This watch tracks federal legislative proposals or executive actions in the US regarding the establishment of an AI Sovereign Wealth Fund to give Americans an equity stake in AI advancements.") has fired! Let's see: the watch is `ongoing` because it tracks "federal legislative proposals or executive actions in the US regarding the establishment of an AI Sovereign Wealth Fund". Since it is ongoing, we should record an event but set `resolves: false` to keep tracking it as the bills actually move through Congress or get signed. Let's call `record_event` on the watch `US AI Sovereign Wealth Fund Legislative or Executive Actions` (ID `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`). Let's read if there is an existing note about this. Yes, there is: `us-government-proposes-equity-stakes-in-ai-labs` ("Trump Administration and Lawmakers Explore Direct Government Equity Stakes in Leading AI Labs Amid Nationalization Threats"). Let's update `us-government-proposes-equity-stakes-in-ai-labs` to cover the latest details: Sanders' "American AI Sovereign Wealth Fund Act" proposal (50% tax) and Trump's "American sovereign wealth fund" musings ( modeled on the 9.9% Intel deal), and how both are aligning on government equity stakes in AI. Let's write the note first. This is an incredibly powerful policy update.

This is excellent! The note has been updated successfully. Now let's record the event on the watch `US AI Sovereign Wealth Fund Legislative or Executive Actions` (ID `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`). Since this is an ongoing watch tracking federal legislative proposals or executive actions, we will keep it open (`resolves: false`). Let's call `record_event`.

Awesome! The event was successfully recorded. Now let's resolve the thread associated with the watch `OpenAI and Anthropic Custom Chip Milestones` using `resolve_thread`. Wait, let's call `resolve_thread` on thread `019eab3a-9c09-7440-a735-3a275cee1f9a`. Let's do that now.

This is wonderful! The thread is now resolved. Let's see: are there any other threads we should resolve? No, the other open threads are `Apple CEO Transition to John Ternus`, `Argentina Non-Human Corporation Legislation Vote`, and `Meta Muse Spark Developer API Release` (which are all ongoing or have future target dates). Wait! Let's check if we should register any forward-looking watches using `record_watch`. Yes! We have several forward-looking catalysts: 1. **Anthropic's 24-hour update on the Fable 5 shutdown:** Anthropic promised to share more details about the U.S. government's export control directive and their path to compliance within 24 hours (by Saturday, June 13, 2026, or Sunday, June 14, 2026). This is a great `one_shot` dated catalyst to watch for in the next cycle. 2. **Trump's meeting with AI executives:** President Trump has scheduled a meeting with top AI executives (including Sam Altman and Dario Amodei) next week (mid-June 2026) to discuss the details of the U.S. AI Sovereign Wealth Fund and equity-sharing model. This is a brilliant `one_shot` dated catalyst. 3. **OpenAI's next model release on June 22, 2026:** Commenters on Hacker News mentioned that OpenAI's next model release is expected on June 22, 2026, as an answer to Mythos. This is a very interesting `one_shot` dated catalyst. Let's register these three watches using `record_watch` to make sure future cycles track them. First, let's register the watch for Anthropic's 24-hour update on Fable 5. Let's call `record_watch`.

Excellent! The watch has been registered successfully. Now let's register the watch for Trump's meeting with AI executives next week. Let's call `record_watch`.

Wonderful! The watch has been registered successfully. Now let's register the watch for OpenAI's next model release on June 22, 2026. Let's call `record_watch`.

Superb! All three watches have been registered successfully. Let's double-check if there are any other open threads or actions needed. No, we have successfully: 1. Listed the open threads and watches. 2. Conducted thorough research on fresh, recent AI developments (the Anthropic Fable 5/Mythos 5 global shutdown, the SpaceX/xAI compute deals with Google and Anthropic, Meta's Muse Spark API delay, and the U.S. AI Sovereign Wealth Fund proposals). 3. Updated the living notes using stable slugs: - `anthropic-expands-project-glasswing-claude-mythos` (Fable 5 / Mythos 5 sudden shutdown) - `meta-muse-spark-api-delays` (Meta Muse Spark API delay) - `clive-chan-exits-openai-joins-anthropic` (OpenAI's Titan custom chip tape-out and Chan's defection) - `us-government-proposes-equity-stakes-in-ai-labs` (Trump and Sanders' AI Sovereign Wealth Fund proposals) 4. Recorded events on active watches (OpenAI custom chip milestones and the Sovereign Wealth Fund). 5. Resolved the OpenAI custom chip milestone thread. 6. Registered three new forward-looking watches (Anthropic's 24-hour update, Trump's executive meeting, and OpenAI's rumored June 22 model release). This has been an incredibly productive and high-impact research cycle. I am ready to call `finish_cycle` and conclude the cycle. Let's do that now.

I have successfully completed the research cycle for topic `019e92c9-99b4-7b6c-bb81-1e0494672f70`. Below is a summary of the findings and actions taken during this cycle. ### Summary of Major Findings 1. **U.S. Export Controls Force Global Claude Fable 5 & Mythos 5 Shutdown** (Slug: `anthropic-expands-project-glasswing-claude-mythos`) * **The Event:** On Friday, June 12, 2026—just three days after launch—Anthropic was forced to abruptly disable access to its newest flagship models, **Claude Fable 5** and **Claude Mythos 5**, for all customers globally. The shutdown was triggered by a U.S. Commerce Department export control directive ordering Anthropic to suspend access to these models by any "foreign national" (including Anthropic's own international engineers, H-1B visa holders, and green card holders). * **The Backlash:** Developers who piped Fable 5 into production are scrambling, and the sudden lockout has provoked severe developer backlash. Closed-source U.S. models are now viewed as a major "supply chain risk," driving international enterprise interest toward sovereign AI initiatives (like France's Mistral) and Chinese open-weight models (like DeepSeek v4 Pro, Kimi K2.7, and MiniMax M3). * **Pretext:** The government cited national security concerns over a shown "jailbreak" used to identify simple software vulnerabilities. Anthropic has publicly pushed back, stating the vulnerabilities are minor and can be discovered by other public models (including OpenAI's GPT-5.5) without any bypass. 2. **OpenAI's "Titan" Custom Chip Reaches TSMC Tape-Out** (Slug: `clive-chan-exits-openai-joins-anthropic`) * **The Milestone:** OpenAI's first custom AI accelerator, codenamed **"Titan"**—designed under Project Nexus in collaboration with Broadcom, Marvell, and TSMC—has officially reached the critical milestone of tape-out. The 3nm chip is designed to reduce OpenAI's reliance on Nvidia's costly GPUs and is slated for mass production in H2 2026. * **Infrastructure Defection:** The milestone comes as Clive Chan, OpenAI's custom silicon leader and second-ever hardware employee, defected to Anthropic to lead their custom silicon esfuerzos, bringing OpenAI's design-to-tape-out playbooks to their chief rival. 3. **Bipartisan Alignment on a U.S. AI Sovereign Wealth Fund** (Slug: `us-government-proposes-equity-stakes-in-ai-labs`) * **Bernie Sanders' Bill:** Senator Bernie Sanders (I-VT) published an op-ed proposing his **American AI Sovereign Wealth Fund Act**, which would impose a one-time 50% tax on large AI companies, converting that tax into public equity to give Americans a direct stake in AI wealth. * **Trump's "Intel Model" Proposal:** President Trump has proposed a voluntary equity-swap model modeled on the government's recent 9.9% equity stake in Intel. Under Trump's model, AI labs like OpenAI would "donate" shares in exchange for federal support and access to government compute. * **Pre-Release Audits:** This sits alongside Trump's newly signed Executive Order, **"Promoting Advanced Artificial Intelligence Innovation and Security"** (NSPM-11), which mandates that frontier labs grant federal national security agencies full access to their models for safety testing at least 30 days prior to public release. 4. **Meta Delays Muse Spark Developer API Again** (Slug: `meta-muse-spark-api-delays`) * **The Delay:** Meta has repeatedly postponed the developer API rollout for **Muse Spark**, its flagship closed-source AI model first unveiled in April 2026. The API has faced multiple delays due to testing bugs and infrastructure bottlenecks, raising investor concerns regarding Meta's ability to monetize its projected $145 billion AI capex. Meta is currently targeting a June 2026 launch window. --- ### Thread & Watch Status * **Resolved Thread:** * `OpenAI and Anthropic Custom Chip Milestones` (ID: `019eab3a-9c09-7440-a735-3a275cee1f9a`): Resolved following OpenAI's "Titan" chip reaching tape-out and Clive Chan's defection to Anthropic. * **Ongoing Watch Event Recorded:** * `US AI Sovereign Wealth Fund Legislative or Executive Actions` (ID: `019eab3a-8c4e-7d5b-b7a4-e1e84554cfe8`): Event recorded covering Trump and Sanders' separate AI Sovereign Wealth Fund proposals (Cato Institute & NYT). * **Registered New Forward-Looking Watches:** * `Anthropic Update on Fable 5 and Mythos 5 Shutdown` (Fire date: June 14, 2026): Tracks Anthropic's promised 24-hour update on the U.S. government's export control directive. * `Trump Meeting with AI Executives on Sovereign Wealth Fund` (Fire date: June 18, 2026): Tracks Trump's upcoming meeting with AI chiefs to discuss giving Americans a direct stake in AI companies. * `OpenAI Rumored Next Model Release` (Fire date: June 22, 2026): Tracks OpenAI's rumored competitive response to Claude Mythos 5.