More than two dozen technology companies, including Microsoft, Meta, Palantir, and IBM, have issued an open letter to policymakers advocating for the widespread adoption and support of open-source artificial intelligence (AI) systems and code. The letter, posted on Friday, argues that an open-source approach is inherently safer and more beneficial for societal innovation than restricting access or relying solely on a limited number of proprietary, closed models.
The signatories contend that the current debate around AI mirrors the software industry's landscape in the 1980s, when concerns about open-source software's impact on business proved unfounded. That era ultimately led to a robust open-source ecosystem that now forms the backbone of the internet, government IT, and commercial software, fostering a shared knowledge base for future innovation. The companies assert that U.S. leadership in AI will be defined by its ability to build a strong, open ecosystem that permeates all sectors, rather than by the success of a single frontier AI model.
While acknowledging the security risks associated with expanding access to open-source AI, the letter posits that the benefits for cybersecurity defenders outweigh the potential for misuse by attackers. Cybersecurity experts have warned that broadly available AI tools could empower less technically skilled criminals. Once an AI model is "open weight," meaning its internal parameters are accessible, anyone can download, customize, and remove built-in safeguards for their own purposes, potentially leading to an increase in malicious deepfakes and other AI-generated content.
However, the letter argues that open-source AI models are crucial for startups, universities, research labs, and other smaller organizations, enabling them to innovate and refine the technology for broader societal utility. The companies state that open weights allow organizations to select the most appropriate model for specific tasks at an optimal cost, reserving high-capacity frontier models for complex problems and deploying efficient, specialized models for everyday applications. This disciplined approach, they claim, is essential for making AI economically sustainable as its use expands.
From a cybersecurity perspective, the letter emphasizes that open-source AI will empower defenders to outpace attackers. It argues that in an environment where attackers leverage advanced AI, defenders require access to comparable models to effectively detect, simulate, and respond to emerging threats. Open models, the signatories contend, enhance defensive capabilities, increase transparency, and facilitate the discovery and remediation of vulnerabilities across multiple teams.
Other notable signatories include Perplexity, Mistral, NVIDIA, Mozilla, The Linux Foundation, Hugging Face, and Dell Technologies. The letter emerges as U.S. policymakers continue to navigate the complex balance between fostering domestic AI innovation and implementing oversight and regulation to mitigate potential harms.
The Trump administration has explored various frameworks, shifting from a laissez-faire stance to an executive order establishing a voluntary industry testing regime. More recently, the administration imposed export controls on Anthropic's Fable model and reportedly urged OpenAI to delay model releases due to cybersecurity concerns. There have also been reports that the administration is considering an executive order to restrict American access to Chinese-made open-source AI models.
The White House, while recognizing the overall advantages of an open-source approach, is also cautious about actions that could inadvertently benefit rival nations. Earlier this month, the White House announced the creation of the Gold Eagle AI cybersecurity clearinghouse, an initiative aimed at coordinating efforts among government, the private sector, and civil society to identify and address AI-discovered vulnerabilities. A senior White House official confirmed that supporting providers and maintainers of open-source AI tools is a significant component of this initiative.
