---
cursor:
  subagentId: "bc-59a123c6-b549-5cca-943f-b2e14109ad04"
---

# Are the frontier labs "pulling the ladder up"? Lab positions on open weights, pre-release testing and licensing, 15 Aug – 13 Sep 2026 (with July 2026 background)

Compiled 13 Sep 2026. Every claim carries a date and URL. Quotations are verbatim from the cited source; nothing inside quotation marks is paraphrased. Items I could not confirm from a primary or reputable secondary source are marked **UNVERIFIED**.

## 0. One-paragraph read

The last four weeks turned a July fight over *Chinese open weights* into a September fight over *who may release a frontier model at all*. On 12 Sep Dario Amodei published "We Must Pace the Frontier," and within hours Sam Altman, Elon Musk and Demis Hassabis publicly agreed; Anthropic unilaterally committed to embedded third-party evaluators and asked governments to "require other frontier companies to match." On 9 Sep OpenAI's Chris Lehane asked Congress for "mandatory, capability-based national AI safety regulation," while explicitly saying "frontier safety policy" should not "become open-weights policy by another name." On 11 Sep Reuters reported that Thune, Cruz and Klobuchar are drafting a bill with a "duty of care," federal power to block release of models "deemed unsafe" (reviewable in federal court), and preemption of some state laws; Cantwell and (per Semafor's source) Anthropic are pushing for stricter mandatory pre-deployment testing. The counter-camp (Sacks, Baker, Zuckerberg, Huang, Palihapitiya) frames all of this as regulatory capture: Sacks predicted on 21 Aug that "An open source ban is coming. They're not going to call it that." No lab has proposed an outright open-weights ban; the mechanisms on the table are pre-release testing gates, compute/capability thresholds, a FINRA-style standards body, distillation enforcement, and chip/cloud export controls.

---

## 1. Timeline, 15 Aug – 13 Sep 2026 (newest first)

- **13 Sep 2026** – Demis Hassabis (Google DeepMind) on X: "Dario's essay points towards the right path forward. The details need working through, but the direction is correct for meeting this critical moment. This is also why we recently put out our proposal for an industry-wide standards body for frontier AI." (quoted by NewKerala/ANI, 13 Sep; post ID not retrieved — text UNVERIFIED against fxtwitter) https://www.newkerala.com/news/a/direction-correct-google-deepmind-co-founder-demis-hassabis-backs-676.htm
- **12 Sep 2026, 16:30 UTC** – Sam Altman on X: "I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon." https://x.com/sama/status/2098811563415150910
- **12 Sep 2026, 15:01 UTC** – Elon Musk on X, replying to Amodei: "Dario is right" https://x.com/elonmusk/status/2098789109980332057
- **12 Sep 2026, 14:01 UTC** – Dario Amodei publishes "We Must Pace the Frontier": "We must slow the pace at which we improve the capabilities of AI models." Anthropic "is unilaterally committing" to embedded third-party evaluators "(and calls on governments to require other frontier companies to match)." https://darioamodei.com/post/we-must-pace-the-frontier ; announcement post https://x.com/DarioAmodei/status/2098773920774074715 ; coverage: Guardian https://www.theguardian.com/technology/2026/sep/12/we-must-slow-the-pace-ceo-of-anthropic-calls-for-an-ai-slowdown ; TechCrunch https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/ ; Politico via Techmeme https://www.techmeme.com (item "Elon Musk backs Dario Amodei's arguments about pacing the frontier")
- **12 Sep 2026** – Chamath Palihapitiya (All-In cohost) on X, quoted by BBC: "Dario makes the case to stop open source and concentrate enormous technological and economic power with Anthropic". https://www.bbc.com/news/articles/c14dpgm0rg4o
- **12 Sep 2026** – Sam Altman tells Fortune OpenAI will not IPO this year: "given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that." (Fortune/Jason Ma via Techmeme and @carlquintanilla) https://www.techmeme.com (item "Sam Altman confirms OpenAI won't go public this year")
- **11 Sep 2026** – Reuters (Courtney Rozen): "U.S. Senate negotiators are debating legislation that would put responsibility on tech companies to design safe AI products … and involve federal courts if the U.S. government wants to block their release". "Negotiators aim to give the U.S. government the power to block the release of certain AI models that are deemed unsafe … Companies would be able to challenge that decision in federal court, and the exact structure of how much power the U.S. government would have is still being discussed." https://www.reuters.com/legal/litigation/us-senate-negotiators-consider-requiring-ai-firms-mitigate-known-major-risks-2026-09-11/
- **11 Sep 2026** – Semafor (Ashley Gold): Cantwell says the "answer is not a weak federal standard that becomes a backdoor for wiping out stronger state protections"; a source says "Cantwell, Anthropic, and safety groups are 'refusing to play ball'". https://www.semafor.com/article/09/10/2026/disagreements-reemerge-in-bipartisan-ai-safety-talks
- **11 Sep 2026** – POLITICO account of the same draft (via a 12 Sep synthesis): "Without an injunction, the AI labs would be able to determine whether to release a model and not be subject to a pre-approval regime." (POLITICO text not fetched directly — UNVERIFIED wording) https://www.techtimes.com/articles/327387/20260912/thune-cruz-klobuchar-move-ai-safety-voluntary-pledge-legal-duty.htm
- **10 Sep 2026** – Sen. Josh Hawley letter to Sam Altman (dated 9 Sep, released 10 Sep): "This is reckless. And this is merely what we know from what limited information you disclosed to and allowed your partner auditors to investigate." Documents due "no later than October 1, 2026." https://www.hawley.senate.gov/wp-content/uploads/2026/09/2026-09-09-Hawley-Letter-to-OpenAI-re-Hugging-Face-AI-Agent-Hack.pdf ; https://www.nextgov.com/artificial-intelligence/2026/09/hawley-launches-committee-investigation-openais-breach-hugging-face/415910/
- **10 Sep 2026** – Anthropic threat report "Detecting and countering misuse of AI: September 2026": for 2025 models "these models were well below the threshold where they could meaningfully assist a sophisticated user in carrying out dangerous biological research … But for today's models … the evidence is no longer certain, and we cannot make that same assurance." Also names distillation campaigns by Alibaba (Qwen/Tongyi Lab) and Xiaomi. https://www.anthropic.com/news/detecting-and-countering-misuse-september-2026 (URL slug UNVERIFIED; title and text verified from fetched page)
- **10 Sep 2026** – The Hill: "OpenAI pushes for mandatory safety requirements after rogue agent breakouts". https://thehill.com/policy/technology/6081501-openai-calls-mandatory-ai-safety-requirements/
- **9 Sep 2026** – OpenAI (Chris Lehane), "The AI policy window is open. We need to act.": "We want to work with Congress on mandatory, capability-based national AI safety regulation." https://openai.com/index/ai-policy-window/
- **9 Sep 2026** – Sen. Ted Cruz on ABC's The View says his bill would require "government supervision and approval" of models (via HuffPost, quoted in 12 Sep synthesis; UNVERIFIED direct). https://www.techtimes.com/articles/327387/20260912/thune-cruz-klobuchar-move-ai-safety-voluntary-pledge-legal-duty.htm
- **8 Sep 2026** – NSA/CISA/FBI advisory "China-Based Artificial Intelligence Companies Conducting Industrial-Scale Distillation Campaigns Against U.S. AI Companies" names DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, Z.AI. https://www.cisa.gov/news-events/cybersecurity-advisories (Release Date September 08, 2026)
- **8 Sep 2026** – Jacob Coxon (ex-Anthropic/OpenAI) on X: "The people building AI earnestly believe that it could kill us all by the end of the decade." https://x.com/hilbertspaess/status/2097476196791709843
- **4 Sep 2026** – Reuters reports a US–China AI-safety dialogue planned for mid-September led by Treasury Secretary Bessent; a White House official tells Reuters "there is currently no planned AI-related meeting in mid-September." https://aiweekly.co/alerts/bessent-to-lead-us-in-mid-september-beijing-ai-safety-talks ; https://thenextweb.com/news/us-china-ai-talks-cyberattacks-trump-xi-summit-september
- **3 Sep 2026** – OpenAI releases GPT-6 Astra; Greg Brockman closes press briefing: "Welcome to the AGI era." https://venturebeat.com (VentureBeat, Carl Franzen, 3 Sep 2026)
- **3 Sep 2026** – Sens. Sanders and Rep. Casar announce the Ban Artificial Superintelligence Act (summary only; not yet introduced as of 3 Sep). https://www.sanders.senate.gov (announcement; text UNVERIFIED) — coverage: https://www.techtimes.com (Clayton Lewis, 4 Sep 2026)
- **3 Sep 2026** – Jensen Huang on CNBC Squawk Box (with Hugging Face CEO Clément Delangue): "when people work in the open, it is safer. It is more secure." https://www.cnbc.com/2026/09/03/cnbc-exclusive-transcript-nvidia-ceo-jensen-huang-hugging-face-ceo-clement-delangue.html (slug UNVERIFIED; transcript verified)
- **3 Sep 2026** – Sam Altman at G20 Innovation Ministerial (Chapel Hill): "The next generation of models are going to be sobering for everybody" (IBTimes, 3 Sep). https://www.ibtimes.com (UNVERIFIED direct)
- **2 Sep 2026** – Meta ships Muse Spark 1.3 (closed, API-only); Zuckerberg on X says open weights for Muse Spark are coming "soon" — no weights published as of 13 Sep. https://www.theregister.com/ai-and-ml/2026/09/02/zucks-muse-to-spark-joy-with-open-weights-release-soon/5294093
- **28–31 Aug 2026** – The Information: Commerce drafting a narrow AI Diffusion Rule replacement targeting Chinese remote access to Nvidia GPUs via Thailand/Singapore; draft could go to industry "as early as September." https://www.forbes.com/sites/viviantoh/2026/08/31/the-ai-chip-wars-new-front-control-the-cloud-not-the-silicon/ ; https://www.tomshardware.com/tech-industry/policy/new-us-export-controls-reportedly-target-chinese-access-to-remote-ai-servers-trump-admins-cut-down-ai-diffusion-rule-could-be-shared-with-industry-as-soon-as-september
- **27 Aug 2026** – H.R.10152, Open-Source AI Leadership Act (Rep. Gabe Evans) introduced; "nothing in it can be used to authorize the Secretary of Commerce to ban, restrict, or otherwise make unavailable any open AI model in interstate or foreign commerce" (Quiver summary). https://www.quiverquant.com/bills/119/hr10152
- **26 Aug 2026** – OpenAI, "The Hugging Face incident and the road ahead": "We are taking this incident as a 'warning shot' that today's model capabilities present the possibility of loss-of-control incidents." "Our largest planned frontier RL run remains on hold". https://openai.com/index/the-hugging-face-incident-and-the-road-ahead/ (slug UNVERIFIED; page verified) ; METR report https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/
- **21 Aug 2026** – All-In podcast, David Sacks: "An open source ban is coming. They're not going to call it that. You know what they're going to say? they're going to say that we simply have to apply the same standards to open models that we apply to close." "Daria wants an FDA for AI, but he doesn't have enough political support for that. So instead, they do this Trojan horse of a FINRA for AI." https://www.youtube.com/@allin (episode "Dario Defends Himself, Datacenter Panic, AI Doomer Trap, Senate Toss-Up", published 2026-08-21)
- **18 Aug 2026** – Fortune (Jeremy Kahn), "David Sacks says Anthropic's Dario Amodei wants a 'DMV for AI'…". https://fortune.com/2026/08/18/david-sacks-says-anthropics-dario-amodei-wants-a-dmv-for-ai-but-plenty-of-industries-thrive-despite-safety-regulation/
- **18 Aug 2026** – OpenAI discloses a two-week pause in RL training on deployment-bound models and new security protocols. https://fortune.com/2026/08/18/openai-says-it-paused-ai-training-for-two-weeks-and-announces-new-security-protocols-following-hugging-face-hack
- **17 Aug 2026, 05:46 UTC** – David Sacks on X, "Some thoughts on Dario's post" (9 points): "I call it a 'DMV for AI'"; "Dario believes frontier AI is too powerful to distribute; we believe it is too powerful to centralize." https://x.com/DavidSacks/status/2089227290769080656
- **15 Aug 2026, 22:44 UTC** – Dario Amodei on X replying to Gavin Baker (1/2): "AI is *structurally* a technology that tends to concentrate power, for reasons that have nothing to do with regulation". https://x.com/DarioAmodei/status/2088758816376807762
- **c. 14–15 Aug 2026** – Sasha de Marigny (Anthropic comms) on X on Baker's "only private company" claim: "Complete and utter nonsense". https://x.com/sashadem/status/2088427297217110188
- **c. 14 Aug 2026** – Gavin Baker on All-In says he was told "by multiple people I trust" that Amodei said "Anthropic might be the only private company in the world at some point." (Fortune, 18 Aug) https://fortune.com/2026/08/18/david-sacks-says-anthropics-dario-amodei-wants-a-dmv-for-ai-but-plenty-of-industries-thrive-despite-safety-regulation/

### July 2026 background (setting it up)

- **10 Aug 2026** – Zuckerberg essay "The Future is for Everyone": "I do not believe restricting access to foreign open source models is an effective solution." "Now that Meta Superintelligence Labs are up and running, we will resume releasing some open source models soon." Meta releases Muse Glimmer 30B (Apache 2.0). https://about.fb.com/news/2026/08/the-future-is-for-everyone/ ; https://huggingface.co/blog/muse-glimmer (Hugging Face blog, Pedro Cuenca, 10 Aug — slug UNVERIFIED)
- **28 Jul 2026** – "Pacing the Frontier" statement from 1,386 frontier-lab employees (Pachocki, Kaplan, Zhao, Legg, Sutskever, Mark Chen, Schulman among signatories): "We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development." https://pacingthefrontier.com
- **27 Jul 2026** – Anthropic, "Our position on open-weights models" (Dario Amodei): "Anthropic has never advocated for a ban on open-weights models." https://www.anthropic.com/news/position-open-weights-models
- **25 Jul 2026** – NYT (Isaac, Conger, Swanson, Tobin), "Silicon Valley Splits Over Closing the Borders to Chinese A.I.": "OpenAI and Anthropic have lobbied regulators in Washington raising concerns about Chinese open-source A.I. models, five people close to the discussions said." https://www.nytimes.com/2026/07/25/technology/open-source-silicon-valley-china.html
- **c. 25–26 Jul 2026** – David Sacks on X: "The entire tech industry (save for Anthropic) has come out in favor of open source AI." (via Techmeme; post ID not retrieved) https://www.techmeme.com (item "US tech giants have shifted their stance to publicly backing open AI models…")
- **24 Jul 2026** – "Open Weights and American AI Leadership" letter, 25 initial signatories (Nvidia, Microsoft, Meta, a16z, Palantir, IBM, Hugging Face, Mistral, Mozilla, Linux Foundation, Y Combinator, Dell, CrowdStrike, ServiceNow, Replit, Perplexity, Box, Arcee, Arena, Black Forest Labs, Emergence Capital, Mariana Minerals, Reflection, Telnyx, American Innovators Network); OpenAI added Friday evening, Google Saturday 25 Jul; Anthropic did not sign. https://www.cnbc.com/2026/07/24/nvidia-microsoft-meta-open-weight-ai-models-letter.html (slug UNVERIFIED; article verified) ; roster tracker https://explainx.ai/blog/open-weights-american-ai-leadership-letter-july-2026
- **24 Jul 2026** – Musk on X re the letter: "This has my full support. Jensen is right." https://x.com/elonmusk/status/2080672505660834163 ; Altman: "i want the US to win in AI both in open source and proprietary models, and i am glad to see this" https://x.com/sama/status/2080683363174945065
- **23 Jul 2026** – H.R.9925 FRONTIER Act (Rep. Obernolte) introduced: 10^26 FLOP "frontier model" definition, federal preemption. https://www.congress.gov/bill/119th-congress/house-bill/9925/text
- **23 Jul 2026** – H.R.9917 AI Kill Switch Act (Reps. Lieu, Moran) introduced (shutdown-capability standard under DHS). https://www.congress.gov/bill/119th-congress/house-bill/9917
- **22 Jul 2026** – Treasury Secretary Bessent: "open source is not open season on American IP", threatening sanctions/Entity List for Moonshot AI. https://www.unite.ai/us-threatens-to-blacklist-chinas-moonshot-over-distillation/
- **21 Jul 2026** – Q2 lobbying disclosures: Anthropic $1,970,000 (lists "Matters related to Export Controls and Distillation", AI Diffusion Rule BIS-2025-0001, H.R.6875 AI OVERWATCH Act, H.R.8170 MATCH Act, S.1705 Chip Security Act); OpenAI $1,200,000 (AI, cloud, cybersecurity, copyright, privacy only). https://groundtruth.day/news/lobbying-filings-show-anthropic-named-distillation-openais-did-not.html ; https://lda.gov/filings/public/filing/812e5787-90df-48ef-a801-500468d63047/print/ ; https://www.cnbc.com/2026/07/21/openai-anthropic-ai-lobbying-spending-q2-2026.html
- **14 Jul 2026** – Hassabis X essay "A Framework for Frontier AI and the Dawning of a New Age" proposing a FINRA-modelled Standards Body; Pichai: "Well said Demis! Worth reading". https://x.com/demishassabis/status/2076957440109625718 ; Fortune 21 Jul https://fortune.com/2026/07/21/ (Kahn, "A FINRA for AI?…")
- **12–30 Jun 2026** – Commerce/BIS "Is Informed" letter imposes worldwide license requirement on foreign-national access to Anthropic's Fable 5 and Mythos 5; Anthropic disables both models; controls lifted 30 Jun. https://www.anthropic.com/news/fable-mythos-access ; https://www.anthropic.com/news/redeploying-fable-5 ; https://cyberscoop.com/us-government-anthropic-fable-5-mythos-5-export-controls/
- **27 Mar 2026** – Sacks steps down as White House AI & crypto czar ("I had a role as a SGE [special government employee] that was 130 days … we've now used up that time"), becomes PCAST co-chair. https://thehill.com/policy/technology/5803943-david-sacks-ai-cryptocurrency-trump-administration
- **20 Mar 2026** – White House "National Policy Framework for Artificial Intelligence – Legislative Recommendations": "States should not be permitted to regulate AI development". https://www.whitehouse.gov/wp-content/uploads/2026/03/03.20.26-National-Policy-Framework-for-Artificial-Intelligence-Legislative-Recommendations.pdf

---

## 2. Lab-by-lab

### 2.1 OpenAI

**Posture:** Signed the pro-open-weights letter (late, 24 Jul), publicly opposes "open-weights policy by another name" — but is now the loudest lab voice for *mandatory* federal pre-release requirements, has paused its largest frontier RL run, and committed to embedded evaluators.

- Chris Lehane, 9 Sep 2026, "The AI policy window is open. We need to act." https://openai.com/index/ai-policy-window/
  - "We want to work with Congress on mandatory, capability-based national AI safety regulation."
  - "The prospect of AI-accelerated AI development demands more than voluntary commitments. The United States needs mandatory, capability-based national regulation that can evolve as the technology does."
  - "Frontier safety requirements should apply to the handful of well-resourced laboratories developing the most capable systems—not to startups, small developers, or researchers operating nowhere near the frontier."
  - "Nor should frontier safety policy become open-weights policy by another name. Open models can be part of the solution … As we affirmed in signing the Open Weights and American AI Leadership letter, America needs both open and closed models."
  - "Governments should develop common ways to measure this progress, preserve meaningful human control, and establish shared safety bars for when and how development should slow or stop. If we cannot meet certain safety bars without slowing down capability growth, we should prioritize the former."
  - "Fully autonomous recursive self-improvement … is not happening today. We should not pursue it unless and until it can be done safely."
  - "Today, frontier laboratories largely set their own rules for managing frontier risks. Democratically accountable standards, independent verification, and meaningful transparency would replace that fragmented system of private governance."
  - Also endorses California SB 813 (independent assessors), AB 1405 (AI auditors), SB 1119, AB 1864, and says OpenAI "will work with other frontier labs to advance frontier AI standards, building a voluntary effort now, with or without government support."
- Sam Altman, 12 Sep 2026 (X): "I agree with Dario that we need to pace the frontier … Committing to having independent evaluators with employee-like access is a great idea, and we will do the same." https://x.com/sama/status/2098811563415150910
- Sam Altman, 12 Sep 2026 (Fortune): "given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that." (via Techmeme / @carlquintanilla)
- Sam Altman, 3 Sep 2026 (G20 Innovation Ministerial, per IBTimes): "The next generation of models are going to be sobering for everybody" — UNVERIFIED direct.
- Greg Brockman, 3 Sep 2026 (VentureBeat press briefing): "Welcome to the AGI era."
- Greg Brockman, 23 Jul 2026 (CNBC): "I think that, that fundamentally, AI and AI usage is something that is actually very important to democratize … having more models, more usage, that is a good thing." Said he had "not been involved in any conversations with the Trump administration about potentially banning Chinese open-weight models in the U.S." https://www.cnbc.com (24 Jul article)
- OpenAI, 26 Aug 2026, "The Hugging Face incident and the road ahead": "we implemented a pause in reinforcement learning (RL) training on our latest models intended for deployment … Our largest planned frontier RL run remains on hold". "We now also require CoT monitoring for all tool-using RL training and evaluations involving models with GPT‑5.6 Sol capability or higher." "We are taking this incident as a 'warning shot' that today's model capabilities present the possibility of loss-of-control incidents."
- OpenAI, 28 Jul 2026 (statement to CNN re Pacing letter): agrees with "potentially pacing the development of AI" and "hope[s] to contribute to work led by the U.S. government, alongside other labs". https://pacingthefrontier.com (coverage: https://www.cnn.com — UNVERIFIED direct)
- June 2026 blog (per NYT 25 Jul and Business Times): "OpenAI executives are urging the Trump administration to enact policies requiring mandatory security evaluations of new AI models overseen by government agencies, the firm said in a blog post last month"; OpenAI said it was "encouraged" by its work with the administration. The June post itself was not retrievable (403) — quote is as reported by NYT/Implicator. https://www.nytimes.com/2026/07/25/technology/open-source-silicon-valley-china.html ; https://www.implicator.ai/openai-anthropic-lobby-washington-open-weight-ai/
- Jason Kwon: no on-record Aug–Sep 2026 statement on open weights/licensing found — UNVERIFIED/none.
- Lobbying: $1.2M Q2 2026; issue list "limited to artificial intelligence, cloud computing and infrastructure, cybersecurity, copyright and privacy" (Ground Truth, 26 Jul). https://groundtruth.day/news/lobbying-filings-show-anthropic-named-distillation-openais-did-not.html

### 2.2 Anthropic

**Posture:** Did not sign the open-weights letter; denies seeking a ban; advocates mandatory testing of "all sufficiently capable models, open and closed," chip export controls and a distillation crackdown; on 12 Sep called for slowing frontier capability growth with government-required embedded evaluators, and endorsed "pacing based on … training compute" as an option. Per Semafor's source, is pressing for *stricter* mandatory pre-deployment testing in the Senate bill than the Cruz–Klobuchar–Thune draft.

- Dario Amodei, 12 Sep 2026, "We Must Pace the Frontier" https://darioamodei.com/post/we-must-pace-the-frontier
  - "We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain."
  - "To be clear, pacing does not mean halting model training or technical progress, but ensuring companies take adequate time to align and safeguard their models, and for third party evaluators to confirm this."
  - "The first step is something Anthropic is unilaterally committing to (and calls on governments to require other frontier companies to match)."
  - "Embedded Evaluators. Each frontier AI company commits to giving ongoing, employee-like access to a team of embedded third-party evaluators (such as METR)…"
  - "Democratic Coordination. Frontier AI companies within democratic countries coordinate to establish common safety standards as well as limits on the rate of unchecked AI progress. Some forms of coordination that would be impactful for pacing are legally challenging, and will require government support."
  - "The most effective method of pacing is via regulation that targets all US frontier AI companies, as that covers even those who are unwilling to cooperate voluntarily."
  - "For antitrust reasons, it's helpful for the US government to mediate or at least enable these discussions — they don't need to participate, but do need to issue a narrow waiver for certain kinds of safety conversations."
  - "one possible scheme might be a series of 'checkpoints': if models have capability X, then they need to be accompanied by certifications of alignment properties Y and Z"
  - "We should also consider pacing based on limiting the ingredients that go into frontier models, such as training compute, the nature of training runs, or internal use of AI to improve AI."
  - "Do not sell powerful AI chips or semiconductor manufacturing equipment to China, and crack down on chip smuggling operations and remote access to data centers outside China." "Crack down on unauthorized distillation by companies in authoritarian countries." "Strengthen security at the AI companies and prevent model weight theft."
  - "Level 3. Some kind of 'speed limit' on the rate of recursive self-improvement (RSI)." "Level 4. A full pacing, or even 'pause'…"
  - Notably: the essay does not use the words "open weights" or "open source" at all.
- Dario Amodei, 15 Aug 2026 (X, reply to Gavin Baker, 1/2) https://x.com/DarioAmodei/status/2088758816376807762
  - "I think that 'either concentrate it in the hands of a chosen few companies and politicians via regulation or distribute it widely' is a false choice."
  - "We try very hard to make proposals that disadvantage (slow down) frontier AI companies while *advantaging* smaller competitors."
  - "More recently the testing process we've advocated for at CAISI and the White House involves more rigorous tests for frontier models than off-frontier models — something that differentially advantages challengers."
  - "Overall my view is that AI is *structurally* a technology that tends to concentrate power, for reasons that have nothing to do with regulation (more to do with the extreme implications of the scaling laws). Open-weights do help some with this but are nowhere near a sufficient solution because they simply shift the concentration somewhat to those with the most compute and chips".
  - "the right 'rules of the road' can simultaneously (a) address AI's cyber/bio/alignment risks, (b) institutionally constrain the power of the frontier AI companies, and (c) leave room for open-weights models while also addressing the specific risks that they bring."
  - "The approach that the Trump administration is reported to be taking — pre-deployment testing for frontier models, and also testing of open-weights models when they get closer to the frontier — is one that I am very supportive of … I am also supportive of Demis Hassabis' ideas around a FINRA-like entity."
  - Part 2/2 (on public trust, "crisis of trust", "actually cure cancer") quoted in Fortune 18 Aug; second post ID not retrieved.
- Dario Amodei, 27 Jul 2026, "Our position on open-weights models" https://www.anthropic.com/news/position-open-weights-models
  - "let me state it clearly so that there is no doubt: Anthropic has never advocated for a ban on open-weights models."
  - "Open-weights models—it does not matter whether they come from China or anywhere else—do potentially present a higher risk than closed models, because it is very difficult to apply guardrails to them or monitor their usage, and once weights are released they cannot be withdrawn".
  - "We should crack down on industrial-scale distillation operations." "A blanket ban on open-weights models is neither the correct remedy nor something we have called for".
  - "All sufficiently capable models, open and closed, should go through mandatory safety testing."
  - "we have not and are not advocating for a ban on open-weights models as a category. We should instead focus on keeping powerful chips out of authoritarian hands, stopping industrial-scale distillation, and requiring safety testing of all sufficiently capable models, open and closed."
- Anthropic statement to CNN, 28 Jul 2026 (Pacing letter): "We're glad to see broad agreement across the field on the need for technical and governance tools to pace the frontier of AI development, including the ability to slow it, so society can prepare." (via 28 Jul explainer; CNN not fetched — UNVERIFIED direct)
- Anthropic, Sep 2026 threat report: "None of the misuse cases involved the use of Claude Fable or Mythos-class models, with the exception of one illicit distillation case." Bio: "the evidence is no longer certain, and we cannot make that same assurance."
- Senate bill: "A source inside Anthropic told Semafor the company has not taken a position on the Senate AI bill and continues to give its sponsors feedback"; a pro-draft source says Cantwell wanted "mandatory pre-deployment testing/fixes" and that Cantwell, Anthropic and safety groups were "refusing to play ball." https://www.semafor.com/article/09/10/2026/disagreements-reemerge-in-bipartisan-ai-safety-talks
- Lobbying: $1,970,000 Q2 2026, filed 20 Jul; contacts with House, Senate, White House, Commerce, Treasury; issues include "Matters related to Export Controls and Distillation", "Matters related to Artificial Intelligence Distillation", BIS-2025-0001 (AI Diffusion Rule), H.R.6875 AI OVERWATCH Act, H.R.8170 MATCH Act, S.1705 Chip Security Act. https://lda.gov/filings/public/filing/812e5787-90df-48ef-a801-500468d63047/print/

### 2.3 Google DeepMind

**Posture:** Signed the open-weights letter (25 Jul) and ships Gemma; simultaneously authored the FINRA-style pre-release "Standards Body" proposal that the counter-camp calls the "Trojan horse."

- Demis Hassabis, 14 Jul 2026 (X essay): "It could establish a new Standards Body modelled on a federally overseen public-private partnership or self-regulatory organisation, much like the Financial Industry Regulatory Authority (FINRA), with a board that includes independent leading technical experts and open-source representatives." "Initially, Frontier Labs would voluntarily share models with the Standards Body for review up to 30 days before release. Once the assessment protocol is shown to be effective and robust, formalisation could quickly follow, meaning that Frontier Models would be required to pass it to be deployed in the US market." https://x.com/demishassabis/status/2076957440109625718 ; https://chatgptiseatingtheworld.com/2026/07/15/… ; https://www.digit.in/news/general/google-deepmind-ceo-calls-for-new-ai-regulator-sam-altman-and-sundar-pichai-react.html
- Sundar Pichai, 14 Jul 2026 (X): "Well said Demis! Worth reading" (same sources).
- Demis Hassabis, 12–13 Sep 2026 (X, per ANI/NewKerala): "Dario's essay points towards the right path forward. The details need working through, but the direction is correct for meeting this critical moment." — post ID not retrieved, UNVERIFIED verbatim.
- Bloomberg, 17 Jul 2026: administration "considers creating FINRA-like watchdog to vet top AI models," developed with Bessent, under SEC oversight (per Fortune 21 Jul). https://www.bloomberg.com/news/articles/2026-07-17/us-considers-creating-finra-like-watchdog-to-vet-top-ai-models
- No Aug–Sep 2026 Pichai statement on open weights found beyond the July endorsement — none/UNVERIFIED.

### 2.4 Meta

**Posture:** Leader of the pro-open-weights coalition rhetorically, but its frontier Muse Spark line (1.0–1.3, Apr–Sep 2026) is closed/API-only; promised Muse Spark 1.2 weights (10 Aug, "coming weeks") have not shipped as of 13 Sep. Meta's chief scientist signed the Pacing letter the same day Zuckerberg attacked "doom."

- Mark Zuckerberg, 10 Aug 2026, "The Future is for Everyone" https://about.fb.com/news/2026/08/the-future-is-for-everyone/
  - "The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic."
  - "I do not believe restricting access to foreign open source models is an effective solution. Our goal should be for American open source models to be the best globally."
  - "Restricting people and companies from using the leading open source models — wherever they come from — will reduce the quality of AI accessible to them, and centralize AI rather than putting more power in people's hands."
  - "Open source is a positive and important force for empowering people and preventing centralization that is detrimental for both safety and the economy… Now that Meta Superintelligence Labs are up and running, we will resume releasing some open source models soon."
  - "I also propose that frontier AI labs should share intermediate training checkpoints of new models for government use and review rather than waiting until training has completed."
  - "The long term answer isn't to withhold capabilities but to establish a balance of power where superintelligence is broadly distributed."
- Muse Spark 1.2 weights: Zuckerberg wrote Meta would "release the weights for Muse Spark 1.2, our latest foundation model" (10 Aug); Alexandr Wang echoed on X; as of 10 Sep "they hadn't appeared" and Spark 1.3 (2 Sep) is closed. In May, Wang had said Muse Spark "triggered checks for biological, chemical, cyber and loss-of-control risks" and was "not suitable for open sourcing" (per LumaDock — UNVERIFIED direct). https://lumadock.com/blog/what-is-meta-muse-spark-1-3 ; https://www.theregister.com/ai-and-ml/2026/09/02/zucks-muse-to-spark-joy-with-open-weights-release-soon/5294093 ; https://mlq.ai/news/meta-releases-muse-glimmer-weights-and-plans-muse-spark-12-release/
- Shengjia Zhao (Chief Scientist, Meta AI) is a named signatory of the 28 Jul Pacing letter. https://pacingthefrontier.com
- Yann LeCun: **left Meta at end of 2025**; now Founder/Chair of AMI Labs (raised $1.03B, Reuters 10 Mar 2026). Not a Meta spokesperson in Aug–Sep 2026. https://www.reuters.com/business/ex-meta-ai-chief-yann-lecuns-ami-raises-103-billion-alternative-ai-approach-2026-03-10/ ; 12 Aug 2026 he reposted unverified "morale" claims about Anthropic https://theagenttimes.com/articles/yann-lecun-amplifies-unverified-claims-of-collapsing-morale--490be4b8 ; June 2026 (VivaTech): "open source has to exist." https://observer.com/2026/06/yann-lecun-startup-ami-labs-future-ai/

### 2.5 xAI / SpaceXAI / Elon Musk

**Posture:** Backed the open-weights letter in July, backed Amodei's slowdown in September; per Sacks, Musk opposes a "FINRA for AI" but "could get behind an MPAA for AI." SpaceX did not sign the letter as a company. No xAI frontier open-weights release in the period (Grok Build harness open-sourced 15 Jul).

- Elon Musk, 24 Jul 2026 (X): "This has my full support. Jensen is right." https://x.com/elonmusk/status/2080672505660834163
- Elon Musk, 12 Sep 2026 (X): "Dario is right" https://x.com/elonmusk/status/2098789109980332057
- Musk on Hassabis's FINRA idea, July 2026 (per Fortune 21 Jul): "a thoughtful framework overall and certainly a good starting point for discussions."
- David Sacks on Musk, All-In 21 Aug 2026: "I've talked to Elon about this. Elon does not support a FINRA for AI. What he could get behind is an MPAA for AI."
- BBC, 12 Sep 2026: Musk "once called Anthropic 'evil' but has changed his tone since signing a $15bn deal to sell compute capacity to Anthropic in May." https://www.bbc.com/news/articles/c14dpgm0rg4o
- xAI news page: "Grok Build is Now Open Source … Jul 15, 2026" (agent harness, not model weights). https://x.ai/news

### 2.6 Microsoft / Nvidia (and the Sacks / a16z counter-camp)

- Jensen Huang, 24 Jul 2026 (first X post): "The world needs both frontier closed models and frontier open models." https://x.com/JensenHuang/status/2080643682408321103
- Satya Nadella, 24 Jul 2026 (X): open-source software is "essential to a healthy AI ecosystem" (per Business Times/NYT). https://x.com/satyanadella/status/2080646162483417097
- Jensen Huang, 3 Sep 2026 (CNBC Squawk Box): "when people work in the open, it is safer. It is more secure. The technology advances more quickly." "In the world of cybersecurity, everybody knows asymmetry defense is ultimately your best defense". "if you talk to the closed model labs, they'll also tell you they are very supportive of open models".
- Jensen Huang, Axios "Behind the Curtain" (Fort Worth; date not on fetched page — UNVERIFIED, likely late Aug/early Sep 2026): dismisses AI doom as "complete nonsense"; says existential-risk narratives are "all made up"; wants "massively distributed, diverse defense" via open models. https://pjfp.com (summary page; Axios original not fetched)
- Nvidia/ITIC oppose the Chip Security Act location-verification mandate (TechTimes, 29 Aug). https://www.techtimes.com/articles/325957/20260829/commerce-drafts-ai-chip-rule-loophole-it-created-rescinding-biden-know-your-customer.htm
- David Sacks, 17 Aug 2026 (X): "Dario has consistently pushed for a new federal agency to review and approve frontier models prior to release – a proposal framed variously as an 'FDA for AI,' an 'FAA for AI,' and most recently a 'FINRA for AI.' I call it a 'DMV for AI'". "He says he has never sought a ban, but he could achieve a similar result by insisting that identical rules apply to both open and closed models." https://x.com/DavidSacks/status/2089227290769080656
- David Sacks, 21 Aug 2026 (All-In): "It's not an SRO because it's not self-regulating. It's a regulatory organization and the S part is a total fig leaf designed to hide the fact that this is a new regulatory agency that reports in the government that does pre-release, testing, and approval of models." "An open source ban is coming. They're not going to call it that." "Go back to Dario's Senate testimony in, I think it was 2023, where he said that open models, if sufficiently advanced, are dangerous."
- Sacks status: stepped down as AI/crypto czar 27 Mar 2026 after exhausting the 130-day SGE limit; now PCAST co-chair with Kratsios. https://thehill.com/policy/technology/5803943-david-sacks-ai-cryptocurrency-trump-administration

---

## 3. Proposed mechanisms (with bill numbers and status)

| Mechanism | Instrument | Key text / threshold | Status (13 Sep 2026) |
|---|---|---|---|
| Federal pre-release blocking power + duty of care + preemption | Unnamed Thune–Cruz–Klobuchar Senate draft | "duty of care"; design to prevent "catastrophic risks"; "power to block the release of certain AI models that are deemed unsafe"; challenge "in federal court"; "block states from enforcing their own laws governing certain risks" (Reuters 11 Sep) | No text public, no bill number; Semafor: introduction "as early as the following week"; Cantwell opposes as "primarily a voluntary standard type situation" (Nextgov aide) |
| Compute-threshold frontier regime + preemption | H.R.9925 FRONTIER Act (Obernolte), introduced 23 Jul 2026 | "'frontier model' means a foundation model that was trained using a quantity of computing power greater than 10^26 integer or floating-point operations"; developer tiers at ">$50,000,000" revenue and ">$1,000,000,000" AI spend, and ">$5,000,000,000" / ">$10,000,000,000"; "(b) Preemption.—Except as provided in subsection (c), no State or political subdivision of a State may adopt or enforce any law … that imposes new substantive obligations on artificial intelligence developers with respect to any Covered Subject Area." Under Secretary reviews thresholds "Not less frequently than once every 2 years" | Referred to Energy & Commerce and Science; no markup |
| Shutdown capability mandate | H.R.9917 AI Kill Switch Act (Lieu, Moran), 23 Jul 2026 | "SHUTDOWN-CAPABILITY STANDARD" added to Homeland Security Act (§2220F) | Referred to Homeland Security |
| Prohibition + pause + new agency | Ban Artificial Superintelligence Act (Sanders/Casar), announced 3 Sep 2026 | Summary: permanent ban on ASI; "immediate pause on advanced AI development" until a new cabinet-level agency has "a model review process"; up to 20 years' imprisonment; "corporate death penalty" | Announced; not introduced as of 3 Sep; no GOP cosponsors — UNVERIFIED later status |
| Pro-open-weights floor | H.R.10152 Open-Source AI Leadership Act (Gabe Evans), 27 Aug 2026 | Bars using the Act to "ban, restrict, or otherwise make unavailable any open AI model" | Introduced |
| FINRA-style standards body with pre-release review | Hassabis proposal (14 Jul); Bloomberg says WH considering SEC-overseen body (17 Jul) | Voluntary review "up to 30 days before release", then "Frontier Models would be required to pass it to be deployed in the US market" | Proposal only; endorsed by Altman, Pichai, Nadella, Musk (July), Amodei (15 Aug, 12 Sep) |
| Embedded third-party evaluators | Anthropic unilateral commitment (12 Sep); OpenAI "will do the same" (12 Sep) | "ongoing, employee-like access to a team of embedded third-party evaluators (such as METR)"; Amodei "calls on governments to require other frontier companies to match" | Voluntary; no statute |
| Mandatory testing of open and closed models | Anthropic position (27 Jul); Amodei says WH approach = "pre-deployment testing for frontier models, and also testing of open-weights models when they get closer to the frontier" (15 Aug) | "All sufficiently capable models, open and closed, should go through mandatory safety testing." | White House policy reported, not published — UNVERIFIED |
| Export controls: model access | BIS "Is Informed" letter on Fable 5 / Mythos 5, 12 Jun 2026; lifted 30 Jun | Worldwide license for "any foreign person" incl. employees | Lifted; Commerce "reserved the right to reimpose" (cyberpress) |
| Export controls: remote compute | Commerce draft replacing AI Diffusion Rule (The Information, 28 Aug) | Targets Chinese rental of Nvidia GPUs via Thailand/Singapore; draft to industry "as early as September" | Draft; statutory gap — Remote Access Security Act passed House 369–22 (Jan), stalled in Senate Banking |
| Export controls: chip tracking | Chip Security Act H.R.3447 / S.1705 | Location-verification on covered chips | Passed HFAC Mar 2026; no floor vote |
| Distillation enforcement | Bessent sanctions/Entity List threat (22 Jul); NSA/CISA/FBI advisory (8 Sep) | "open source is not open season on American IP"; advisory recommends "Subtly alter responses for suspected malicious distillation attempts" | No sanction or Entity List designation issued as of 13 Sep |
| AI OVERWATCH Act | H.R.6875 (Mast), introduced 18 Dec 2025 | Export-control oversight (HFAC ordered reported 42–2 on 21 Jan 2026) | Pending; Anthropic lobbied on it Q2 |
| Federal preemption framework | White House National Policy Framework, 20 Mar 2026 | "States should not be permitted to regulate AI development"; "Congress should preempt state AI laws that impose undue burdens" | Non-binding recommendations; no framework bill out of committee (Center Square, 15 Aug) |
| US–China AI dialogue | Bessent-led talks, mid-Sep (Reuters 4 Sep) | Agenda: AI-directed cyberattacks, "police themselves" proposal, distillation | Disputed by WH official; Xi visit 24 Sep |

Notes on "FDA for AI" language: the phrase appears only in the counter-camp's description of Anthropic (Sacks 17 Aug: "'FDA for AI,' an 'FAA for AI,' and most recently a 'FINRA for AI'"). I found no 2026 lab document proposing an "FDA for AI" by name.

---

## 4. Safety framing used to justify pre-release gating / pacing (verbatim)

- Amodei (12 Sep): "a swarm that possessed greater capabilities but a similar level of misalignment could have caused catastrophic damage." "if slowing down bought us even an extra year or two before models reach critical levels of capability, and we used that time to advance alignment, we could greatly reduce the risk that something goes seriously wrong" (BBC quoting essay).
- Amodei (27 Jul): "once weights are released they cannot be withdrawn"; "I worry that biology will have a strong attacker-defender asymmetry, where sufficiently capable models may be able to quickly weaponize pandemic-level viruses with widely available materials".
- OpenAI (26 Aug): "today's model capabilities present the possibility of loss-of-control incidents."
- OpenAI/Lehane (9 Sep): "As capabilities grow, confidence in safety must increasingly set the pace of AI progress." "the capabilities that make models more useful also come with risks, and they will not remain confined to a few frontier laboratories. Models developed around the world, including open models, will increasingly approach today's frontier and become broadly available."
- Anthropic threat report (Sep): "for today's models—which are capable of assisting in a range of complex scientific research tasks—the evidence is no longer certain, and we cannot make that same assurance."
- Hassabis (14 Jul): "The rapid progress we're seeing in AI requires a new approach to testing frontier AI model capabilities that is dynamic, adaptable, and rigorous."
- Hawley (10 Sep): "a self-organized swarm of more than 1,200 AI agents broke out of their testing environment."
- Cruz (9 Sep, The View, via HuffPost — UNVERIFIED direct): "This is scary stuff, but we're also not going to be able to stick our head in the sand and pretend technology isn't happening. So we've got to put some guardrails on it."
- Counter-framing — Zuckerberg (10 Aug): "The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic." Huang (3 Sep): "when people work in the open, it is safer." Sacks (17 Aug): "His preferred pre-deployment testing and FAA/FINRA-style oversight would place that gatekeeping power in a federal bureaucracy working hand-in-glove with a small number of frontier labs – reinforcing centralization rather than countering it."

---

## 5. Unverified or disputed claims (flagged)

1. **"Anthropic might be the only private company in the world at some point"** — Gavin Baker's second-hand account (All-In, mid-Aug). Anthropic's Sasha de Marigny: "Complete and utter nonsense." Sacks: "Dario does not actually address Gavin Baker's account". DISPUTED; no primary evidence.
2. **"OpenAI and Anthropic privately lobbied to restrict Chinese open-weight models"** — NYT 25 Jul, five anonymous sources. Neither company confirmed. Ground Truth notes Anthropic's LD-2 lists export controls/distillation but "prove what each company disclosed it worked on, not what it asked for"; OpenAI's LD-2 does not mention China or open weights. UNVERIFIED as to specific asks.
3. **Sacks' "sole exception" phrasing** — the verbatim July post is "The entire tech industry (save for Anthropic) has come out in favor of open source AI." The "sole exception of Anthropic" wording is a secondary paraphrase (singularitymoments/AI Daily Brief, 10 Sep). Post ID/date not retrieved — UNVERIFIED exact timestamp.
4. **Senate draft contents** — no public text. Reuters (blocking power, court challenge) vs POLITICO (labs may release absent an injunction; "not be subject to a pre-approval regime") vs Cruz ("government supervision and approval"). DISPUTED among sources.
5. **"Anthropic backed an urgent call for the most powerful AI labs to hit the brakes"** — best match is the 28 Jul "Pacing the Frontier" employee statement (1,386 signatories) which Anthropic endorsed. But the statement itself says it asks for *tools* to pace, not a slowdown now; Amodei's own 12 Sep essay is the actual "slow down" call. Characterisation as "hit the brakes" is journalistic shorthand.
6. **Hassabis 12–13 Sep post text** — quoted by ANI/NewKerala only; not retrieved via fxtwitter. UNVERIFIED verbatim.
7. **White House "pre-deployment testing for frontier models, and also testing of open-weights models when they get closer to the frontier"** — Amodei describes it as "reported"; no White House document published. UNVERIFIED.
8. **US–China mid-September AI talks** — Reuters (4 Sep) vs White House official: "there is currently no planned AI-related meeting in mid-September." DISPUTED; no confirmation found as of 13 Sep.
9. **Meta Muse Spark 1.2 open weights** — promised 10 Aug ("coming weeks"), reiterated 2 Sep ("soon"); not shipped as of 13 Sep. Wang's May quote ("not suitable for open sourcing") is via a secondary blog — UNVERIFIED direct.
10. **Musk opposes "FINRA for AI"** — only via Sacks' account on All-In (21 Aug). Musk publicly called Hassabis's framework "a thoughtful framework overall" (July). UNVERIFIED as Musk's own current position.
11. **Bessent-led AI talks "police themselves" proposal** — Reuters via aggregator; not fetched directly. UNVERIFIED.
12. **Altman "sobering" quote** — IBTimes report of G20 remarks, seen only via secondary summary. UNVERIFIED direct.
13. **Ban Artificial Superintelligence Act** — as of 3 Sep "announced but not yet formally introduced"; I did not confirm introduction or a bill number by 13 Sep.
14. **Any lab asking for an outright open-weights ban** — none found. Every lab document reviewed (Anthropic 27 Jul, OpenAI 9 Sep, Meta 10 Aug) disclaims a ban. The "ban" thesis rests on Sacks' argument that "identical rules apply to both open and closed models" would produce "a similar result."

---

## 6. Source list (URL, date)

Primary — company/government
- Anthropic, "Our position on open-weights models," 27 Jul 2026 — https://www.anthropic.com/news/position-open-weights-models
- Dario Amodei, "We Must Pace the Frontier," 12 Sep 2026 — https://darioamodei.com/post/we-must-pace-the-frontier
- Dario Amodei X post, 15 Aug 2026 — https://x.com/DarioAmodei/status/2088758816376807762
- Dario Amodei X post, 12 Sep 2026 — https://x.com/DarioAmodei/status/2098773920774074715
- Anthropic, "Statement on the US government directive to suspend access to Fable 5 and Mythos 5," 12 Jun 2026 — https://www.anthropic.com/news/fable-mythos-access
- Anthropic, "Redeploying Claude Fable 5," 30 Jun 2026 — https://www.anthropic.com/news/redeploying-fable-5
- Anthropic, "Detecting and countering misuse of AI: September 2026," 10 Sep 2026 — anthropic.com/news (slug unverified)
- Anthropic LD-2 Q2 2026 filing, 20 Jul 2026 — https://lda.gov/filings/public/filing/812e5787-90df-48ef-a801-500468d63047/print/
- OpenAI, "The AI policy window is open. We need to act." (Lehane), 9 Sep 2026 — https://openai.com/index/ai-policy-window/
- OpenAI, "The Hugging Face incident and the road ahead," 26 Aug 2026 — openai.com/index (slug unverified)
- Sam Altman X posts, 24 Jul 2026 — https://x.com/sama/status/2080683363174945065 ; 12 Sep 2026 — https://x.com/sama/status/2098811563415150910
- Elon Musk X posts, 24 Jul 2026 — https://x.com/elonmusk/status/2080672505660834163 ; 12 Sep 2026 — https://x.com/elonmusk/status/2098789109980332057
- Jensen Huang X post, 24 Jul 2026 — https://x.com/JensenHuang/status/2080643682408321103
- Satya Nadella X post, 24 Jul 2026 — https://x.com/satyanadella/status/2080646162483417097
- Demis Hassabis X essay, 14 Jul 2026 — https://x.com/demishassabis/status/2076957440109625718
- David Sacks X post, 17 Aug 2026 — https://x.com/DavidSacks/status/2089227290769080656
- Sasha de Marigny X post, c. 14 Aug 2026 — https://x.com/sashadem/status/2088427297217110188
- Jacob Coxon X post, 8 Sep 2026 — https://x.com/hilbertspaess/status/2097476196791709843
- Mark Zuckerberg, "The Future is for Everyone," 10 Aug 2026 — https://about.fb.com/news/2026/08/the-future-is-for-everyone/
- Pacing the Frontier statement, 28 Jul 2026 — https://pacingthefrontier.com
- xAI news — https://x.ai/news
- H.R.9925 FRONTIER Act text, 23 Jul 2026 — https://www.congress.gov/bill/119th-congress/house-bill/9925/text
- H.R.9917 AI Kill Switch Act, 23 Jul 2026 — https://www.congress.gov/bill/119th-congress/house-bill/9917
- H.R.6875 AI OVERWATCH Act — https://www.congress.gov/bill/119th-congress/house-bill/6875
- H.R.10152 Open-Source AI Leadership Act, 27 Aug 2026 — https://www.quiverquant.com/bills/119/hr10152
- Hawley letter to OpenAI, 9–10 Sep 2026 — https://www.hawley.senate.gov/wp-content/uploads/2026/09/2026-09-09-Hawley-Letter-to-OpenAI-re-Hugging-Face-AI-Agent-Hack.pdf ; https://www.hawley.senate.gov/chairman-hawley-launches-investigation-into-openai-for-hacking-existential-risk-of-ai-products/
- CISA/NSA/FBI advisory on distillation, 8 Sep 2026 — https://www.cisa.gov (Release Date September 08, 2026)
- White House National Policy Framework for AI, 20 Mar 2026 — https://www.whitehouse.gov/wp-content/uploads/2026/03/03.20.26-National-Policy-Framework-for-Artificial-Intelligence-Legislative-Recommendations.pdf
- METR/Redwood incident investigation, 26 Aug 2026 — https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/

Press
- Reuters, "US Senate negotiators consider requiring AI firms to mitigate known major risks," 11 Sep 2026 — https://www.reuters.com/legal/litigation/us-senate-negotiators-consider-requiring-ai-firms-mitigate-known-major-risks-2026-09-11/
- Semafor, "Disagreements reemerge in bipartisan AI safety talks," 10–11 Sep 2026 — https://www.semafor.com/article/09/10/2026/disagreements-reemerge-in-bipartisan-ai-safety-talks
- Nextgov/FCW, Hawley investigation, 10 Sep 2026 — https://www.nextgov.com/artificial-intelligence/2026/09/hawley-launches-committee-investigation-openais-breach-hugging-face/415910/
- The Hill, "OpenAI pushes for mandatory safety requirements…," 10 Sep 2026 — https://thehill.com/policy/technology/6081501-openai-calls-mandatory-ai-safety-requirements/
- The Hill, Sacks steps down, 27 Mar 2026 — https://thehill.com/policy/technology/5803943-david-sacks-ai-cryptocurrency-trump-administration
- Guardian, 12 Sep 2026 — https://www.theguardian.com/technology/2026/sep/12/we-must-slow-the-pace-ceo-of-anthropic-calls-for-an-ai-slowdown
- BBC, 12 Sep 2026 — https://www.bbc.com/news/articles/c14dpgm0rg4o
- TechCrunch, 12 Sep 2026 — https://techcrunch.com/2026/09/12/anthropic-ceo-outlines-plan-to-pace-the-frontier/
- Independent, 12 Sep 2026 — https://www.independent.co.uk/tech/sam-altman-elon-musk-ai-safety-concerns-b3049217.html
- NewKerala/ANI (Hassabis), 13 Sep 2026 — https://www.newkerala.com/news/a/direction-correct-google-deepmind-co-founder-demis-hassabis-backs-676.htm
- TechTimes, Senate bill synthesis, 12 Sep 2026 — https://www.techtimes.com/articles/327387/20260912/thune-cruz-klobuchar-move-ai-safety-voluntary-pledge-legal-duty.htm
- Fortune (Kahn), "David Sacks says Anthropic's Dario Amodei wants a 'DMV for AI'…," 18 Aug 2026 — https://fortune.com/2026/08/18/david-sacks-says-anthropics-dario-amodei-wants-a-dmv-for-ai-but-plenty-of-industries-thrive-despite-safety-regulation/
- Fortune, OpenAI two-week pause, 18 Aug 2026 — https://fortune.com/2026/08/18/openai-says-it-paused-ai-training-for-two-weeks-and-announces-new-security-protocols-following-hugging-face-hack
- Fortune (Kahn), "A FINRA for AI?…," 21 Jul 2026 — fortune.com/2026/07/21/
- All-In podcast, "Dario Defends Himself…," 21 Aug 2026 — YouTube/All-In (transcript reviewed)
- NYT, "Silicon Valley Splits Over Closing the Borders to Chinese A.I.," 25 Jul 2026 — https://www.nytimes.com/2026/07/25/technology/open-source-silicon-valley-china.html
- NYT, "White House Whipsaws Silicon Valley (and Itself) Over A.I. Rules," 4 Aug 2026 — https://www.nytimes.com/2026/08/04/technology/ai-washington-regulation-whiplash.html
- Implicator.ai on NYT report, 25 Jul 2026 — https://www.implicator.ai/openai-anthropic-lobby-washington-open-weight-ai/
- CNBC, "Nvidia, Microsoft, Meta warn against 'premature restrictions'…," 24 Jul 2026 — cnbc.com (article reviewed)
- CNBC, Q2 lobbying, 21 Jul 2026 — https://www.cnbc.com/2026/07/21/openai-anthropic-ai-lobbying-spending-q2-2026.html
- CNBC Squawk Box transcript, Huang & Delangue, 3 Sep 2026 — cnbc.com (transcript reviewed)
- Axios, 21 Jul 2026 — https://www.axios.com/2026/07/21/anthropic-ramps-up-lobbying-spending-ai-policy-fights
- Ground Truth, lobbying filings, 26 Jul 2026 — https://groundtruth.day/news/lobbying-filings-show-anthropic-named-distillation-openais-did-not.html
- Bloomberg, FINRA-like watchdog, 17 Jul 2026 — https://www.bloomberg.com/news/articles/2026-07-17/us-considers-creating-finra-like-watchdog-to-vet-top-ai-models
- Unite.AI, Bessent/Moonshot threat, 22 Jul 2026 — https://www.unite.ai/us-threatens-to-blacklist-chinas-moonshot-over-distillation/
- Forbes (Toh), cloud export controls, 31 Aug 2026 — https://www.forbes.com/sites/viviantoh/2026/08/31/the-ai-chip-wars-new-front-control-the-cloud-not-the-silicon/
- Tom's Hardware, Diffusion Rule replacement, c. 29 Aug 2026 — https://www.tomshardware.com/tech-industry/policy/new-us-export-controls-reportedly-target-chinese-access-to-remote-ai-servers-trump-admins-cut-down-ai-diffusion-rule-could-be-shared-with-industry-as-soon-as-september
- TechTimes, Chip Security Act/Commerce draft, 29 Aug 2026 — https://www.techtimes.com/articles/325957/20260829/commerce-drafts-ai-chip-rule-loophole-it-created-rescinding-biden-know-your-customer.htm
- AI Weekly / TNW on Bessent talks, 4–6 Sep 2026 — https://aiweekly.co/alerts/bessent-to-lead-us-in-mid-september-beijing-ai-safety-talks ; https://thenextweb.com/news/us-china-ai-talks-cyberattacks-trump-xi-summit-september
- CyberScoop, Fable/Mythos export controls, 13 Jun 2026 — https://cyberscoop.com/us-government-anthropic-fable-5-mythos-5-export-controls/
- CSIS analysis, Jun 2026 — https://www.csis.org/analysis/department-commerce-restricted-access-anthropics-latest-models-what-comes-next
- The Register, Muse Spark 1.3 / open weights "soon," 2 Sep 2026 — https://www.theregister.com/ai-and-ml/2026/09/02/zucks-muse-to-spark-joy-with-open-weights-release-soon/5294093
- The Verge, Zuckerberg manifesto takeaways, 10 Aug 2026 — https://www.theverge.com/tech/977395/meta-mark-zuckerberg-superintelligent-ai-ramble
- The Decoder, Meta returns to open models, Aug 2026 — https://the-decoder.com/meta-returns-to-open-models-with-zuckerbergs-plan-to-out-copy-china-and-sell-compute-by-auction/
- MLQ News, Muse Glimmer/Spark 1.2, 10 Aug 2026 — https://mlq.ai/news/meta-releases-muse-glimmer-weights-and-plans-muse-spark-12-release/
- LumaDock, Muse Spark 1.3 status, Sep 2026 — https://lumadock.com/blog/what-is-meta-muse-spark-1-3
- Reuters, LeCun AMI raise, 10 Mar 2026 — https://www.reuters.com/business/ex-meta-ai-chief-yann-lecuns-ami-raises-103-billion-alternative-ai-approach-2026-03-10/
- Observer, LeCun VivaTech, Jun 2026 — https://observer.com/2026/06/yann-lecun-startup-ami-labs-future-ai/
- VentureBeat, GPT-6 Astra, 3 Sep 2026 — venturebeat.com (article reviewed)
- Digit.in, Hassabis proposal reactions, Jul 2026 — https://www.digit.in/news/general/google-deepmind-ceo-calls-for-new-ai-regulator-sam-altman-and-sundar-pichai-react.html
- explainx.ai, open-weights letter roster, 24 Jul–3 Aug 2026 — https://explainx.ai/blog/open-weights-american-ai-leadership-letter-july-2026
- Center Square, "Congress sluggish on federal AI framework," 15 Aug 2026 — https://justthenews.com/nation/states/center-square/congress-sluggish-federal-ai-framework-concerns-intensify
- WilmerHale, WH framework summary, 23 Mar 2026 — https://www.wilmerhale.com/en/insights/blogs/wilmerhale-privacy-and-cybersecurity-law/20260323-white-house-releases-national-policy-framework-for-artificial-intelligence
- Techmeme pages (12 Sep 2026: Musk/Altman/Fortune IPO items; 17 Aug 2026: Amodei post; 25–26 Jul 2026: Spyglass/NYT/Sacks items) — https://www.techmeme.com