The Peterson Institute for International Economics (PIIE) has published a short but sharp assessment of the “Fable–Mythos” episode, a sequence of stop‑gap U.S. controls on AI models, arguing that piecemeal restrictions often backfire and ultimately strengthen Chinese competitors. The PIIE blog post is worth reading in full for its framing and evidence. For broader context on governance debates in this field, see Chatham House’s work on technology policy.
The Fable–Mythos episode
PIIE describes the Fable–Mythos saga as emblematic of a larger pattern: when the United States imposes abrupt restrictions on AI models or model exports without a durable, predictable framework, it creates a set of market and policy responses that undermine U.S. leverage. The most concrete example cited is the short, 18‑day ban on Anthropic activity that PIIE characterizes as backfiring. While the intervention temporarily constrained one company’s activity, it also signaled an ad hoc regulatory approach that can be exploited by well‑capitalized Chinese firms and is supported by state policy aimed at import substitution and indigenous development.
The PIIE note emphasizes two linked dynamics. First, ad hoc controls generate a clear commercial opening: buyers and partners who fear future disruption have stronger incentives to switch to domestic suppliers. Second, when enforcement is inconsistent or narrowly targeted, firms in China can design technical and commercial workarounds — ranging from local fine‑tuning and deployment of open models to routing services through third parties, to neutralize the intended effect of sanctions.
For an example of the broader market competition in AI hardware and software, see EastFrontier’s coverage of the Nvidia–Huawei dynamics and the chip war. The Fable–Mythos episode sits within this wider competitive landscape.
Why ad hoc controls accelerate Chinese substitution
PIIE’s core contention is structural: unpredictable, unilateral controls change incentives in ways that favor China. Several mechanisms are relevant:
• Demand signal: Sudden controls make buyers such as corporations, local governments, and research institutions wary of relying on foreign suppliers that may become unavailable. That risk premium accelerates procurement of domestic alternatives and public investment in homegrown models and inference stack components.
• Supply response: Chinese firms and state planners respond quickly to open market opportunities. When Western vendors are constrained, domestic firms scale existing projects faster, attract talent, and secure financing to capture market share. The result is a more robust indigenous ecosystem that is harder to displace later.
• Circumvention and adaptation: PIIE warns that firms will find ways around narrow or technical bans. These include re‑engineering models to run on different hardware, using third‑country intermediaries, or leaning on open‑source communities and distributed training resources. Those routes reduce the effectiveness of short‑term export controls.
• Political legitimation: Ad hoc rules validate domestic narratives about technological containment, which in turn justify government subsidies, preferential procurement, and regulatory support for national champions.
These dynamics are not hypothetical. The PIIE analysis points to the Anthropic episode as evidence that temporary measures can catalyze the very outcome they were intended to prevent: faster Chinese adaptation and substitution.
For reporting on how China is moving aggressively on domestic AI systems and robotics, areas where substitution matters practically as well as strategically, see EastFrontier’s pieces on China’s humanoid robot boom and its research hubs and on U.S. export controls and regulatory responses.
What this means for U.S. strategy
PIIE’s critique leads to an implicit policy prescription: avoid episodic, unpredictable measures and instead pursue durable, rule‑based approaches that preserve strategic leverage without handing the initiative to rivals. There are three practical takeaways for Washington:
1. Predictability matters. Firms and allied governments need clarity about the scope, duration, and enforcement of controls so commercial choices aren’t driven by fear of sudden cutoffs.
2. Multilateral coordination amplifies effectiveness. Controls agreed with allies reduce the incentive for targeted states to pivot to alternative suppliers and limit opportunities for arbitrage through third countries.
3. Focus on chokepoints rather than one‑off company actions. Measures that target clear, hard‑to‑replicate capabilities (and that can be transparently defended on national security grounds) are less likely to spur quick substitution than blunt, company‑by‑company bans.
PIIE does not claim that no controls are warranted; rather, its point is germane to any policymaker who views technological competition through a cost‑benefit lens. When controls are necessary, their design must account for how market participants, especially those in China’s rapidly evolving tech ecosystem, will rationally respond.
The politics of restraint
Putting these lessons into practice will be politically difficult. Policymakers respond to acute risks and public pressure; ad hoc measures are often tempting because they look decisive. But PIIE’s analysis is a reminder that tactical moves can produce strategic losses. In the U.S.–China high‑tech competition, short‑term signaling can convert into long‑term capability shifts if China’s industrial base reaps the gains.
PIIE’s brief on the Fable–Mythos saga should not be dismissed as technocratic hair‑splitting. It reframes a policy question at the heart of the tech war: should Washington prioritize the immediate appearance of control, or the longer game of denying rivals durable advantages? The evidence from this episode suggests that unless the United States and its partners adopt predictable, narrow, and coordinated measures, ad hoc controls will too often accelerate China’s path to technological autonomy.