A top White House AI advisor warns that American frontier models may be losing the open-source race to China — and their own safety guardrails could be part of the problem.
The a16z Show
A top White House AI advisor warns that American frontier models may be losing the open-source race to China — and their own safety guardrails could be part of the problem.
TL;DR
Sriram Krishnan, fresh off his tenure as Senior White House AI Policy Advisor, joins Theo Jaffee and Sofia Puccini to break down the explosive week in open-source AI — Kimi K3, Qwen, Muse Spark, and more [1] — Sriram Krishnan "Within a matter of days, the AI ecosystem saw Groq 4.5, Muse Spark, Inkling, Kimi K3, and Qwen all ship. Sriram Krishnan calls it a histori…" 03:25 . The surge of capable open-weight models is squeezing frontier lab margins [2] — Sriram Krishnan "NeoCloud & infra layer benefit from open models: As pricing pressure hits frontier labs, NeoCloud providers and infrastructure companies wi…" 06:30 , raising cybersecurity questions about American models being too restrictive while Chinese alternatives face fewer guardrails [3] — Sriram Krishnan "The most capable open-weight models right now are Kimi K3, Deepseek, and Qwen — all Chinese. Sriram Krishnan says flatly: he would much rat…" 15:40 , and creating a distillation arms-race asymmetry that disadvantages U.S. startups. The key takeaway: if an open model provides real value, capitalism will build the supply chain around it.
Sriram Krishnan joins Theo Jaffee and Sofia Puccini just after concluding his tenure as Senior White House Policy Advisor on AI to discuss one of the biggest weeks yet for open-source AI, unpacking the rapid release of models including Kimi K3 and Qwen, pricing pressure on frontier labs, distillation policy, cybersecurity, and U.S.-China AI competition.
The episode opens with a punchy excerpt from Sriram Krishnan making his core argument before the intro rolls: if an open-weight model provides real value, capitalism will build the entire supply chain around it — NeoCloud providers, chip vendors, data center operators and all.
[1] — Sriram Krishnan "If an open-weight model provides real value, every layer beneath it — from NeoCloud providers to chip makers to data center operators — wil…" 19:38The narrator then frames the broader conversation: open-source AI is moving faster than ever, the balance of power in the industry may be shifting, and Krishnan, fresh off his White House tenure, is here to unpack what it all means for frontier labs, AI policy, pricing, cybersecurity, and America's position in the global AI race. It's a tight, effective cold open that signals the episode's agenda. - Theo introduces Sriram Krishnan as someone who needs little introduction: Senior White House AI Policy Advisor, former General Partner at Andreessen Horowitz, and veteran of Microsoft, Meta, Snap, and Twitter. Krishnan, who notes this is his first video appearance without a suit and tie in nearly two years, is in a buoyant mood.
[1] — Sriram Krishnan "It's very clear the Frontier Labs are going to push at the very, very frontier. And so I think they'll continue to be that." 09:09He's a fan of the show, excited to be free of the formality of government service, and ready to engage. The brief exchange sets a relaxed, candid tone for what follows — a conversation between insiders who know the AI industry well. - Theo asks about Krishnan's tweet calling Kimi K3 'a big moment with multiple implications for the entire industry,' and Krishnan delivers.
[1] — Sriram Krishnan "Within a matter of days, the AI ecosystem saw Groq 4.5, Muse Spark, Inkling, Kimi K3, and Qwen all ship. Sriram Krishnan calls it a histori…" 03:25Four or five months ago, he explains, only a handful of frontier models existed at the very top — GPT-5, Opus 4. It felt like the frontier labs were pulling irreversibly away from everyone else. Then, within days: Groq 4.5 from xAI, Muse Spark from Meta, Inkling from Thinking Machines, and finally Kimi K3 and Qwen. The key implications? First, choice — developers can now route their agents and harnesses to multiple capable models. Second, and more troubling: some American frontier models have safety guardrails that block legitimate security research, and researchers are turning to Kimi K3 as a result. Third, pricing pressure on frontier labs is now real — if near-frontier capability is available in open-weight models, token prices will have to drop, compressing frontier lab margins.[2] — Sriram Krishnan "NeoCloud & infra layer benefit from open models: As pricing pressure hits frontier labs, NeoCloud providers and infrastructure companies wi…" 06:30The benefits flow to NeoCloud providers and infrastructure layers, not the labs themselves. - Sofia asks how frontier labs will respond, and Krishnan lays out the dynamics clearly.
[1] — Sriram Krishnan "Sriram Krishnan poses the defining question for frontier AI labs: is their moat in the model itself, or in the product wrapped around it? C…" 10:40Labs like Anthropic and OpenAI will keep pushing at the very frontier — the jagged edge of capability where only they operate — but for a large class of tasks, near-frontier open-weight models will suffice. Email-checking agents, calendar-scanning, routine coding tasks: these don't need GPT-5-level intelligence. This will push pricing pressure onto the frontier labs for their non-frontline offerings. The evidence is already arriving: Anthropic has extended Claude's availability beyond its originally scheduled end, almost certainly in response to competitive pressure from open-weight alternatives.[2] — Sriram Krishnan "Anthropic extended Claude availability: Anthropic extended the availability of Claude (Fable) beyond its originally planned end date, likel…" 10:00The deeper strategic question Krishnan poses: is the moat in the intelligence, or in the harness? Claude Code, Codex, and similar developer products are already remarkable and deeply sticky. If raw AI capability becomes a commodity, the product layer may be what frontier labs actually monetize. - Theo raises an Axios report that the Trump administration is considering restricting Chinese open-source AI models, citing either national security concerns or the fact that they have reached near-Claude-level capability in agentic coding. Krishnan has no inside information from his former colleagues, but he can offer context.
[1] — Sriram Krishnan "The most capable open-weight models right now are Kimi K3, Deepseek, and Qwen — all Chinese. Sriram Krishnan says flatly: he would much rat…" 15:40The Trump administration's AI action plan, which he helped craft, explicitly champions open source in its opening section — so any blanket crackdown on open-source AI would be in tension with the administration's own stated policy. But Krishnan is blunt on one point: the fact that the leading open-weight models — Kimi K3, Deepseek, Qwen — are Chinese rather than American is not a comfortable situation. He rattles off the American efforts underway: Gemma from Google, Nimotron from NVIDIA, Thinking Machines, new startups. He expects them to improve. But the current moment is one where Chinese models lead open source, and that has real implications for national security, economic competitiveness, and the deployment of AI by allied governments. - Krishnan grew up on open source and is a genuine believer in its security advantages.
[1] — Sriram Krishnan "Linus Torvalds said 'given enough eyes, all bugs are shallow.' Sriram Krishnan applies this to AI: open-weight models downloaded from Huggi…" 17:30He invokes Linus's Law — 'given enough eyes, all bugs are shallow' — to argue that open-weight models are more secure than closed ones because anyone can download them from Hugging Face and inspect every layer. You simply cannot do that with a closed API model. But he immediately counterbalances this with a sobering real-time data point: on the day of the recording, Hugging Face reported an active incident where an AI LLM agent was being used to systematically probe and attack their systems. The uncomfortable irony: the answer to AI-powered cyber attacks is almost certainly AI-powered defense — which requires defenders to have access to the best models. If American frontier model safety guardrails make those models less useful for security work, defenders are already at a disadvantage.[2] — Sriram Krishnan "Hugging Face attacked by AI agent: At the time of recording, Hugging Face reported an active incident where an AI LLM agent was being used …" 18:30 - The distillation debate is one of the episode's richest segments. Krishnan starts by resetting the framing: distillation — learning from the outputs of other models — has been foundational to AI development since the beginning.
[1] — Sriram Krishnan "Chinese AI companies can distill reasoning traces from American models with little consequence. Meanwhile, a Silicon Valley startup trying …" 22:10Every model from the original GPT to Claude bootstrapped itself from human knowledge crawled off the internet. And now much of that internet is itself AI-generated (AI slop), which feeds back into training. Krishnan amusingly reveals he once ran his own tweet through an AI detector to confirm it read as human-written. The real issue, he argues, isn't distillation per se — it's the asymmetry. Chinese AI developers can freely harvest reasoning traces from American frontier models, while U.S. startups face genuine legal uncertainty about whether they can distill from other American models.[2] — Sriram Krishnan "Uneven distillation playing field for U.S. vs. China: Chinese AI developers can freely distill reasoning traces from American models, while…" 22:10He credits Sequoia's Dean Meyer for articulating this clearly in a recent post, and Ben Thompson of Stratechery for reaching a similar conclusion independently. The solution: enshrine distillation rights for American model developers to level the playing field. The irony he notes in passing: current American open-weight models themselves use Chinese models as teacher models in fine-tuning, highlighting just how entangled global AI development has become. - The distillation debate is one of the episode's richest segments. Krishnan starts by resetting the framing: distillation — learning from the outputs of other models — has been foundational to AI development since the beginning.
[1] — Sriram Krishnan "Chinese AI companies can distill reasoning traces from American models with little consequence. Meanwhile, a Silicon Valley startup trying …" 22:10Every model from the original GPT to Claude bootstrapped itself from human knowledge crawled off the internet. And now much of that internet is itself AI-generated (AI slop), which feeds back into training. Krishnan amusingly reveals he once ran his own tweet through an AI detector to confirm it read as human-written. The real issue, he argues, isn't distillation per se — it's the asymmetry. Chinese AI developers can freely harvest reasoning traces from American frontier models, while U.S. startups face genuine legal uncertainty about whether they can distill from other American models.[2] — Sriram Krishnan "Uneven distillation playing field for U.S. vs. China: Chinese AI developers can freely distill reasoning traces from American models, while…" 22:10He credits Sequoia's Dean Meyer for articulating this clearly in a recent post, and Ben Thompson of Stratechery for reaching a similar conclusion independently. The solution: enshrine distillation rights for American model developers to level the playing field. The irony he notes in passing: current American open-weight models themselves use Chinese models as teacher models in fine-tuning, highlighting just how entangled global AI development has become.
- Open-weight model
- An AI model whose trained weights are publicly released, allowing anyone to download, inspect, run, and fine-tune it — distinct from fully 'open source' (which includes training code and data) and fully 'closed' (API-only) models.
- Distillation
- A training technique where a smaller or newer AI model learns by studying the outputs (often called 'reasoning traces') of a larger or more capable model, rather than training purely on raw human data.
- Reasoning traces
- The step-by-step chain-of-thought outputs produced by AI models during inference; often used as training data to teach newer models how to reason through problems.
- NeoCloud
- Next-generation cloud inference providers (e.g., Bastion, Fireworks) that specialize in hosting and serving AI model inference at scale, often as an alternative to hyperscalers like AWS or Azure.
- SOTA
- State of the Art — the best currently known performance on a given AI benchmark or task.
- KYC
- Know Your Customer — a compliance process requiring businesses to verify the identity of their users; here applied to AI labs verifying who is accessing their model APIs to prevent abuse.
- Linus's Law
- The principle coined by Linus Torvalds (creator of Linux) that 'given enough eyes, all bugs are shallow' — meaning open community inspection leads to more reliable, secure software.
- Harness
- In AI product contexts, the application layer or developer tooling built around a base model — e.g., Claude Code or Codex — that determines how the model's intelligence is accessed and used.
- Frontier lab
- An AI research organization (e.g., Anthropic, OpenAI, Google DeepMind) working at the absolute cutting edge of model capability, typically requiring massive compute investment.
- Inference cloud
- A cloud service that hosts trained AI models and serves their outputs (inferences) on demand, as distinct from services that host model training.
- Agentic coding
- The use of AI agents — systems that autonomously plan, execute, and iterate — to write, test, and debug code with minimal human intervention.
- RSI
- Recursive Self-Improvement — a hypothesized AI capability where a model can meaningfully improve its own architecture or training, potentially triggering rapid capability gains.
- AI action plan
- A formal policy document released by the Trump administration in early 2025 outlining the U.S. government's priorities and strategy for AI development and governance.
- Fine-tuning
- The process of further training a pre-trained AI model on a smaller, specialized dataset to adapt it for a particular task, domain, or client need.
- AI slop
- Informal term for low-quality, formulaic, or generic AI-generated content that floods the internet — used here to describe how AI-generated text is increasingly mixed into training data.
Chapter 1 · 00:00
Intro & Episode Overview #
The episode opens with a punchy excerpt from Sriram Krishnan making his core argument before the intro rolls: if an open-weight model provides real value, capitalism will build the entire supply chain around it — NeoCloud providers, chip vendors, data center operators and all. [1] — Sriram Krishnan "If an open-weight model provides real value, every layer beneath it — from NeoCloud providers to chip makers to data center operators — wil…" 19:38 The narrator then frames the broader conversation: open-source AI is moving faster than ever, the balance of power in the industry may be shifting, and Krishnan, fresh off his White House tenure, is here to unpack what it all means for frontier labs, AI policy, pricing, cybersecurity, and America's position in the global AI race. It's a tight, effective cold open that signals the episode's agenda.
Chapter 3 · 03:25
The Open-Source AI Explosion: Groq, Muse Spark, Kimi K3, Qwen #
Theo asks about Krishnan's tweet calling Kimi K3 'a big moment with multiple implications for the entire industry,' and Krishnan delivers. [1] — Sriram Krishnan "Within a matter of days, the AI ecosystem saw Groq 4.5, Muse Spark, Inkling, Kimi K3, and Qwen all ship. Sriram Krishnan calls it a histori…" 03:25 Four or five months ago, he explains, only a handful of frontier models existed at the very top — GPT-5, Opus 4. It felt like the frontier labs were pulling irreversibly away from everyone else. Then, within days: Groq 4.5 from xAI, Muse Spark from Meta, Inkling from Thinking Machines, and finally Kimi K3 and Qwen. The key implications? First, choice — developers can now route their agents and harnesses to multiple capable models. Second, and more troubling: some American frontier models have safety guardrails that block legitimate security research, and researchers are turning to Kimi K3 as a result. Third, pricing pressure on frontier labs is now real — if near-frontier capability is available in open-weight models, token prices will have to drop, compressing frontier lab margins. [2] — Sriram Krishnan "NeoCloud & infra layer benefit from open models: As pricing pressure hits frontier labs, NeoCloud providers and infrastructure companies wi…" 06:30 The benefits flow to NeoCloud providers and infrastructure layers, not the labs themselves.
The Open-Source AI Explosion: Kimi K3, Qwen, and What It Means Within a matter of days, the AI ecosystem saw Groq 4.5, Muse Spark, Inkling, Kimi K3, and Qwen all ship. Sriram Krishnan calls it a historic inflection: for the first time in months, the frontier labs no longer have a monopoly on top-tier model capability.
Open-source AI surge: 5+ major models in days Within days, the AI ecosystem saw Groq 4.5, Muse Spark, Inkling, Kimi K3, and Qwen all release, representing an unprecedented burst of open-weight model choice.
American Security Research Is Routing Through Chinese AI U.S. security researchers are using Kimi K3 for legitimate security work because Claude's refusals block them. A former White House AI advisor calls this 'a weird spot to be' — and a genuine national security concern.
U.S. security work routed through Chinese AI A security researcher was using Kimi K3 for security work instead of Claude because Claude's safety refusals blocked legitimate security research tasks.
Open Models Will Erode Frontier Lab Gross Margins When near-frontier open-weight models are available for free or cheap, frontier labs will have to cut token prices to compete. The economics don't disappear — they just shift from frontier labs to NeoCloud providers and infrastructure players.
Frontier labs face token pricing pressure The availability of near-frontier open-weight models will likely force frontier labs to drop token prices, eroding their gross margins.
NeoCloud & infra layer benefit from open models As pricing pressure hits frontier labs, NeoCloud providers and infrastructure companies with GPUs and power stand to capture the shifting economics.
Chapter 4 · 08:50
How Will Frontier Labs Respond? Pricing, Moats, and Claude's Extension #
Sofia asks how frontier labs will respond, and Krishnan lays out the dynamics clearly. [1] — Sriram Krishnan "Sriram Krishnan poses the defining question for frontier AI labs: is their moat in the model itself, or in the product wrapped around it? C…" 10:40 Labs like Anthropic and OpenAI will keep pushing at the very frontier — the jagged edge of capability where only they operate — but for a large class of tasks, near-frontier open-weight models will suffice. Email-checking agents, calendar-scanning, routine coding tasks: these don't need GPT-5-level intelligence. This will push pricing pressure onto the frontier labs for their non-frontline offerings. The evidence is already arriving: Anthropic has extended Claude's availability beyond its originally scheduled end, almost certainly in response to competitive pressure from open-weight alternatives. [2] — Sriram Krishnan "Anthropic extended Claude availability: Anthropic extended the availability of Claude (Fable) beyond its originally planned end date, likel…" 10:00 The deeper strategic question Krishnan poses: is the moat in the intelligence, or in the harness? Claude Code, Codex, and similar developer products are already remarkable and deeply sticky. If raw AI capability becomes a commodity, the product layer may be what frontier labs actually monetize.
Anthropic extended Claude availability Anthropic extended the availability of Claude (Fable) beyond its originally planned end date, likely in response to competitive pressure from open-weight models.
The Real Moat: Intelligence or the Harness? Sriram Krishnan poses the defining question for frontier AI labs: is their moat in the model itself, or in the product wrapped around it? Claude Code, Codex, and similar harnesses may be what keeps frontier labs viable as raw intelligence commoditizes.
Chapter 5 · 13:40
Should the U.S. Government Restrict Chinese Open-Source Models? #
Theo raises an Axios report that the Trump administration is considering restricting Chinese open-source AI models, citing either national security concerns or the fact that they have reached near-Claude-level capability in agentic coding. Krishnan has no inside information from his former colleagues, but he can offer context. [1] — Sriram Krishnan "The most capable open-weight models right now are Kimi K3, Deepseek, and Qwen — all Chinese. Sriram Krishnan says flatly: he would much rat…" 15:40 The Trump administration's AI action plan, which he helped craft, explicitly champions open source in its opening section — so any blanket crackdown on open-source AI would be in tension with the administration's own stated policy. But Krishnan is blunt on one point: the fact that the leading open-weight models — Kimi K3, Deepseek, Qwen — are Chinese rather than American is not a comfortable situation. He rattles off the American efforts underway: Gemma from Google, Nimotron from NVIDIA, Thinking Machines, new startups. He expects them to improve. But the current moment is one where Chinese models lead open source, and that has real implications for national security, economic competitiveness, and the deployment of AI by allied governments.
Trump AI action plan backs open source The Trump administration's AI action plan explicitly highlighted the importance of open source AI in its opening section, signaling early policy support.
America Is Losing the Open-Source AI Race to China The most capable open-weight models right now are Kimi K3, Deepseek, and Qwen — all Chinese. Sriram Krishnan says flatly: he would much rather the leading models be American. Gemma, Nimotron, and new startups are trying, but the U.S. is currently behind.
Chapter 6 · 17:30
Open-Weight Models & Security: Linus's Law Meets the AI Age #
Krishnan grew up on open source and is a genuine believer in its security advantages. [1] — Sriram Krishnan "Linus Torvalds said 'given enough eyes, all bugs are shallow.' Sriram Krishnan applies this to AI: open-weight models downloaded from Huggi…" 17:30 He invokes Linus's Law — 'given enough eyes, all bugs are shallow' — to argue that open-weight models are more secure than closed ones because anyone can download them from Hugging Face and inspect every layer. You simply cannot do that with a closed API model. But he immediately counterbalances this with a sobering real-time data point: on the day of the recording, Hugging Face reported an active incident where an AI LLM agent was being used to systematically probe and attack their systems. The uncomfortable irony: the answer to AI-powered cyber attacks is almost certainly AI-powered defense — which requires defenders to have access to the best models. If American frontier model safety guardrails make those models less useful for security work, defenders are already at a disadvantage. [2] — Sriram Krishnan "Hugging Face attacked by AI agent: At the time of recording, Hugging Face reported an active incident where an AI LLM agent was being used …" 18:30
Linus's Law Applied to AI: Open Weight = Inherently Secure Linus Torvalds said 'given enough eyes, all bugs are shallow.' Sriram Krishnan applies this to AI: open-weight models downloaded from Hugging Face can be inspected by the entire world, making them inherently more auditable than closed frontier models. Security through transparency, not obscurity.
Open-weight models inherently more inspectable Open-weight models downloaded from Hugging Face can be inspected, fine-tuned, and modified by the entire global community, making them inherently more auditable than closed models.
Hugging Face attacked by AI agent At the time of recording, Hugging Face reported an active incident where an AI LLM agent was being used to probe and attempt to breach their systems.
Distillation Is Not a Scandal — It's How AI Has Always Worked
Every AI model from the original GPT onward was trained by distilling human knowledge from the internet. The controversy around Chinese models distilling from American ones ignores the fact that distillation is foundational to how all models are built — the real issue is the asymmetric legal playing field.
Chapter 7 · 19:25
The Distillation Debate: AI Slop, Reasoning Traces, and the Asymmetric Playing Field #
The distillation debate is one of the episode's richest segments. Krishnan starts by resetting the framing: distillation — learning from the outputs of other models — has been foundational to AI development since the beginning. [1] — Sriram Krishnan "Chinese AI companies can distill reasoning traces from American models with little consequence. Meanwhile, a Silicon Valley startup trying …" 22:10 Every model from the original GPT to Claude bootstrapped itself from human knowledge crawled off the internet. And now much of that internet is itself AI-generated (AI slop), which feeds back into training. Krishnan amusingly reveals he once ran his own tweet through an AI detector to confirm it read as human-written. The real issue, he argues, isn't distillation per se — it's the asymmetry. Chinese AI developers can freely harvest reasoning traces from American frontier models, while U.S. startups face genuine legal uncertainty about whether they can distill from other American models. [2] — Sriram Krishnan "Uneven distillation playing field for U.S. vs. China: Chinese AI developers can freely distill reasoning traces from American models, while…" 22:10 He credits Sequoia's Dean Meyer for articulating this clearly in a recent post, and Ben Thompson of Stratechery for reaching a similar conclusion independently. The solution: enshrine distillation rights for American model developers to level the playing field. The irony he notes in passing: current American open-weight models themselves use Chinese models as teacher models in fine-tuning, highlighting just how entangled global AI development has become.
Distillation is core to all AI training Every major AI model, from the original GPT to Claude, has been trained through distillation from human-generated internet content, making distillation an inherent part of AI development.
[Capitalism Will Build the Stack Around Open Models](/bit/podbit/10249/)
If an open-weight model provides real value, every layer beneath it — from NeoCloud providers to chip makers to data center operators — will organize around it. Sriram Krishnan says the spectacular growth of inference clouds and the broader ecosystem already proves this out.
[Capitalism self-organizes around valuable open models](/bit/snapshot/11183/)
If an open-weight model provides genuine value, the entire supply chain — from NeoCloud to chip providers to data center operators — will naturally orient itself to support it.
[AI slop proliferating on the internet](/bit/snapshot/11179/)
A significant and growing share of new internet content is AI-generated, which itself feeds back into training future AI models, compounding the distillation dynamic.
What's Next for Sriram Krishnan After the White House After 18 months as the White House's Senior AI Policy Advisor, Sriram Krishnan is focusing on the mission he's been working on in different forms for years: ensuring America and its allies get access to AI at scale through better government-industry cooperation.
Chapter 8 · 21:50
Outro & Disclaimer #
The distillation debate is one of the episode's richest segments. Krishnan starts by resetting the framing: distillation — learning from the outputs of other models — has been foundational to AI development since the beginning. [1] — Sriram Krishnan "Chinese AI companies can distill reasoning traces from American models with little consequence. Meanwhile, a Silicon Valley startup trying …" 22:10 Every model from the original GPT to Claude bootstrapped itself from human knowledge crawled off the internet. And now much of that internet is itself AI-generated (AI slop), which feeds back into training. Krishnan amusingly reveals he once ran his own tweet through an AI detector to confirm it read as human-written. The real issue, he argues, isn't distillation per se — it's the asymmetry. Chinese AI developers can freely harvest reasoning traces from American frontier models, while U.S. startups face genuine legal uncertainty about whether they can distill from other American models. [2] — Sriram Krishnan "Uneven distillation playing field for U.S. vs. China: Chinese AI developers can freely distill reasoning traces from American models, while…" 22:10 He credits Sequoia's Dean Meyer for articulating this clearly in a recent post, and Ben Thompson of Stratechery for reaching a similar conclusion independently. The solution: enshrine distillation rights for American model developers to level the playing field. The irony he notes in passing: current American open-weight models themselves use Chinese models as teacher models in fine-tuning, highlighting just how entangled global AI development has become.
The Distillation Asymmetry: China vs. U.S. Startups Chinese AI companies can distill reasoning traces from American models with little consequence. Meanwhile, a Silicon Valley startup trying to do the same with another American model faces genuine legal uncertainty. Sriram Krishnan, citing Dean Meyer of Sequoia and Ben Thompson, says this uneven playing field needs to be fixed.
Uneven distillation playing field for U.S. vs. China Chinese AI developers can freely distill reasoning traces from American models, while U.S. startups face legal uncertainty about distilling from other American models, creating an asymmetric disadvantage.
No indexed bits in this chapter.
Show stoppers #
## Snapshots ()
## Key Quotes ()
Sign in to keep viewing
Create a free account to keep exploring this episode's insights, snapshots, quotes and claims.