Jacob Gold Urges Anthropic CEO to Back Mandatory Open Weights Law
An open letter argues that requiring public AI models to be open weights would slow the frontier race by dismantling proprietary valuation models.
Jacob Gold has published an open letter to Anthropic CEO Dario Amodei, challenging the executive to advocate for a law requiring all publicly available AI models to be released as open weights. The proposal seeks to shift the regulatory conversation from safety thresholds to the fundamental economic structure of AI development.
Gold’s proposal specifically targets any AI model offered to the public, though he notes that internal or research models would be excluded from the requirement. This challenge comes as a direct response to Amodei's recent post, "We Must Pace the Frontier," in which the CEO committed Anthropic to providing third-party evaluators with permanent, employee-level access to their systems and called for government-mandated industry-wide adoption of similar safety measures.
The Economic Logic of Open Weights
The debate over open weights—releasing the actual parameters of a model—versus closed, API-only access is a central tension in AI safety and competition. Anthropic, a Public Benefit Corporation (PBC), operates under a mission for the "responsible development and maintenance of advanced AI for the long-term benefit of humanity." While Amodei has previously stated that Anthropic does not advocate for a ban on open-weights models and views them as a potential public good, he has also expressed concerns regarding dangerous capabilities and geopolitical risks.
Gold argues that the current regulatory requests from frontier labs, such as compute thresholds and embedded evaluators, inevitably lead to regulatory capture. Instead, he suggests that mandatory open weights would serve as a more effective brake on the industry. According to Gold, such a law would reduce the capital available for future training runs by fundamentally changing the valuation assumptions of frontier model development.
Implications for the AI Race
If a major frontier lab like Anthropic were to support mandatory open weights, it would effectively destroy the "proprietary moat" that currently justifies multi-billion dollar investments. By removing the ability to maintain a secret, high-value asset, the financial incentive for massive, rapid-fire training runs could diminish. This would potentially achieve the slowing of AI progress that Amodei has called for, but through market mechanics rather than government-managed compute limits.
Strategic Tension
It remains to be seen whether Amodei or Anthropic will formally respond to Gold's challenge. While the company has positioned itself as a safety-first alternative to competitors like OpenAI, the tension between its PBC status and the immense capital requirements of frontier models continues to define its strategic positioning. The industry will be watching to see if the call for "pacing the frontier" extends to dismantling the proprietary nature of the models themselves. This intersection of corporate governance and open-source philosophy represents a critical pivot point for the future of AI regulation and the distribution of power among the world's most capable systems.