The White House is preparing to expand its artificial intelligence policy framework to include open-source models for the first time, sources tell WIRED. The move marks a significant shift in the administration's approach to AI governance, bringing transparency to a regulatory dance the government has performed for months - trying to control a technology without appearing to control it. The update could reshape how companies like Meta and others develop and release open-weight models.
The White House is quietly preparing to bring open-source AI models under its regulatory umbrella, a move that sources say will expand the administration's existing AI policy framework in the coming weeks. According to WIRED, the update represents the government's first comprehensive attempt to address open-weight models - AI systems whose parameters are publicly released, allowing anyone to download and modify them.
The timing isn't accidental. While the administration has spent months touting a light-touch approach to AI regulation, the proliferation of powerful open models from Meta, Microsoft, and research institutions has forced a reckoning. These systems offer unprecedented access to cutting-edge AI capabilities, but they also present unique challenges that closed, proprietary models don't.
Unlike OpenAI's GPT-4 or Google's Gemini - which companies can update, restrict, or shut down at will - open models can't be recalled once released. That permanence has created a regulatory headache. The White House has been trying to figure out how to oversee technology it fundamentally can't control, all while avoiding the appearance of stifling innovation.
The updated framework will likely focus on pre-release safety testing and transparency requirements rather than restricting distribution, sources familiar with the discussions indicate. This approach attempts to thread the needle between security concerns and the open-source community's resistance to heavy-handed regulation. But the details remain fluid, and any policy touching open models risks backlash from researchers who view openness as essential to AI safety and progress.
Meta has become the most prominent champion of open-source AI among major tech companies, releasing its Llama model family with minimal restrictions. The company has argued that open development leads to more robust, safer AI systems through community scrutiny - a position that's put it at odds with competitors like OpenAI that favor closed development.
The policy expansion comes as the administration grapples with mounting pressure from both sides. National security officials worry about adversaries fine-tuning open models for malicious purposes, from generating disinformation to designing biological weapons. Meanwhile, AI researchers and civil liberties groups warn that restricting open models would concentrate power among a handful of corporations and hamper academic research.
The regulatory challenge is complicated by the technology's dual-use nature. The same open model that helps a researcher cure diseases could theoretically help bad actors design new pathogens. The same system that democratizes AI access could be weaponized for sophisticated cyberattacks. The White House framework will need to address these risks without crushing the collaborative development that's accelerated AI progress.
What remains unclear is how the administration will enforce any new requirements. Unlike with Microsoft or Google - companies with deep government relationships and regulatory compliance teams - enforcing rules on decentralized open-source projects presents unique difficulties. Once model weights are released online, they spread across countless servers and jurisdictions within hours.
Industry observers expect the policy to include voluntary commitments initially, similar to the White House's existing AI framework that secured pledges from major tech companies. But sources suggest the administration is exploring whether existing export controls or other regulatory authorities could apply mandatory requirements to certain high-capability open models.
The expansion also signals that the era of AI self-regulation is ending. For months, the administration encouraged companies to develop their own safety standards while threatening regulation if they didn't. This update suggests the government is moving from encouragement to enforcement, even as it tries to maintain its innovation-friendly messaging.
The White House's planned expansion into open-source AI regulation represents a pivotal moment in the governance of artificial intelligence. By attempting to bring open models under policy oversight, the administration is acknowledging that the technology has outpaced its regulatory framework. The challenge now is crafting rules that address legitimate security concerns without strangling the collaborative development that's driven AI breakthroughs. How the White House navigates this tension will likely set the template for AI governance globally - and determine whether open-source AI remains a viable alternative to corporate-controlled systems. For developers, researchers, and companies building on open models, the next few weeks could reshape the entire landscape.
