"The United States now faces a similar choice with artificial intelligence," the companies wrote, framing open-source AI as a strategic alternative to concentration in a few closed systems.
Microsoft and more than two dozen tech companies urge an open-source path
Microsoft, alongside more than two dozen tech firms, posted an open letter on Friday pressing policymakers to support "open-source AI systems and code across society." The letter draws a direct parallel to the software debates of the 1980s, arguing that the long-term outcome of broad access was a "vibrant ecosystem" that underpins modern internet infrastructure, government IT and commercial products. "Our AI leadership will be judged not by one frontier AI model, but by whether the United States builds a strong, open ecosystem that diffuses into every sector," the companies wrote.
Signers and signposts: who backed the letter
Other notable companies signing the letter include Meta, Palantir, Perplexity, Mistral, NVIDIA, Mozilla, The Linux Foundation, Hugging Face, Dell Technologies and IBM. The letter urges expanding access and support for open-source AI, asserting that "open weights let every organization match the right model to the right job at the right cost," and reserving "frontier-scale capability for genuine frontier problems."
Cybersecurity trade-offs: defenders, attackers, and low-level criminals
The letter makes an explicit case for security benefits from openness: "In a world where cybersecurity attackers use advanced AI, defenders need access to models with comparable capabilities so they can detect, simulate, and respond to emerging threats," the companies wrote. They argue open models "broaden defensive capability, increase transparency, and allow vulnerabilities to be discovered and remediated across many teams."
Yet the open approach carries risks the letter concedes and security experts warn about. Cybersecurity experts say one of the biggest beneficiaries of broadly available AI tools will be "low-level criminals who until now lacked the technical expertise or resources to launch serious attacks." The letter acknowledges that once a model is released with open weights "anyone can download it, customize it, strip it of any guardrails and use it for their own purposes." As open-source models improve at generating imagery, the source highlights that "the danger of locally-customized CSAM and sexualized deepfakes could also grow."
Policy friction: the Trump administration, export controls, and Gold Eagle
The letter arrives amid active debates inside the federal government about how to balance industrial support and harm mitigation. According to the source, the Trump administration has pursued several frameworks since taking office: an early laissez-faire stance, an executive order creating a voluntary testing regime, the imposition of export controls on Anthropic’s Fable model, and reported pressure on OpenAI to delay model releases out of cybersecurity concerns. The administration has also reportedly considered an executive order that would restrict American access to Chinese-made open-source models.
At the same time, the White House announced the creation of a Gold Eagle AI cybersecurity clearinghouse earlier this month to coordinate government, private sector and civil society work on AI-discovered vulnerabilities. A senior White House official told reporters that a "big part of that effort" is supporting providers and maintainers of open-source AI tools.
What this means for technologists and security teams, policymakers, and startups, universities, and research labs
- Technologists and security teams: Defenders will push for access to capable models so they can "detect, simulate, and respond" to AI-driven threats; the companies argue open models let defenders discover and remediate vulnerabilities across many teams.
- Policymakers and the White House: Officials face a trade-off between supporting an open ecosystem and restricting access for national-security reasons — already reflected in export controls on Anthropic’s Fable model and consideration of limits on Chinese-made models — even as the Gold Eagle clearinghouse plans to support open-source maintainers.
- Startups, universities, and research labs: The letter highlights that open weights disproportionately benefit smaller entities, enabling them to match models to tasks at lower cost and to innovate without depending on a few frontier providers.
The debate the letter surfaces is straightforward in principle but complex in practice. The signatories frame open-source AI as a route to distributed innovation and stronger defensive tooling, while security experts warn that lowered barriers can empower criminals and enable harmful image-based misuse. With the administration simultaneously deploying export controls, weighing restrictions on foreign open-source models, and announcing a Gold Eagle clearinghouse intended in part to support open-source maintainers, the central question is whether federal policy will tilt toward enabling the broad ecosystem the companies describe — or toward tighter controls aimed at immediate security risks.
Original reporting: https://cyberscoop.com/tech-leaders-open-source-ai-cybersecurity/
