AI Export Controls: Dan Sherrard-Smith Questions US Policy on Advanced Models

In a recent LinkedIn post, Dan Sherrard-Smith discusses the sudden offline status of Anthropic’s powerful AI model, Fable, attributing its unavailability not to safety concerns but to political and export-control issues with the US government. Sherrard-Smith highlights the complexities and potential geopolitical influences shaping the accessibility of cutting-edge AI technology.

The Politics Behind AI Model Availability

Sherrard-Smith points out the unusual circumstances surrounding Fable’s disappearance. The model, released just a week prior by Anthropic, was reportedly made unavailable due to an export-control dispute with the US Commerce Department. This action was prompted by a formal letter warning Anthropic against distributing its advanced models, Mythos and Fable, to “foreign persons,” as reported by Bloomberg.

“The most powerful AI model on the planet is offline today. → Not because of safety. Because of politics.”

According to Sherrard-Smith, internal communications at Anthropic suggest employees feel the company is being “unfairly targeted” and “bullied based on bad vibes.” The situation is further complicated by reports from The Washington Post detailing an expansion of companies with access to Mythos, including a South Korean firm with alleged ties to China. This backdrop makes the situation far from a straightforward safety debate.

Navigating the Shifting Landscape of AI Tools

The implications of such political and regulatory hurdles are significant for businesses and individuals relying on advanced AI. Sherrard-Smith offers a pragmatic approach for navigating this volatile environment:

• Treat AI intelligence as borrowed.
• Use the best model while it’s available.
• Save the plans, docs, and implementation notes.
• Then use them later with a lower-cost or more ethical alternative.

As Dan Sherrard-Smith notes, the broader adoption of AI chatbots by the public presents a contrast to the industry’s focus on specific, high-end models. He highlights that while the industry might be debating models like Claude, the majority of users are engaging with more widely accessible platforms.

“Your buyers are on ChatGPT and Gemini. Our industry is arguing about Claude.”

This observation suggests a potential disconnect between the AI models being developed and regulated at the highest levels and the tools that are actually reaching and being utilized by the general consumer and business user. Sherrard-Smith’s analysis underscores the need for a flexible strategy when integrating AI, acknowledging that the most powerful tools may not always be the most stable or accessible long-term.

Broader Implications for AI Adoption

The situation with Fable, as detailed by Sherrard-Smith, raises critical questions about the future of AI development and international cooperation. The involvement of government bodies in dictating the distribution of AI models, even for non-safety reasons, sets a precedent that could impact innovation and global access to technology.

“This isn’t a safety dispute. Anthropic’s own employees frame it as a relationship issue. Which means the timeline is unpredictable.”

Dan Sherrard-Smith’s post serves as a timely reminder for industry leaders to remain adaptable. While global discussions, like those involving Anthropic’s CEO at the G7, aim to establish regulatory frameworks, the practical realities of AI deployment are subject to rapidly evolving political and economic factors. His advice to leverage available tools, document processes, and plan for future transitions with alternative models offers a grounded perspective amidst the uncertainties of the AI landscape.

Ultimately, Sherrard-Smith concludes with a call for further discussion, asking, “What did I miss? Let me know in the comments.” This invites a broader conversation on the complex interplay between technological advancement, political interests, and market adoption in the realm of artificial intelligence.