On April 16, I responded to the Department of Justice and Federal Trade Commission’s 2026 Joint Public Inquiry on updating the 2000 Collaboration Guidelines. Given major shifts in the global economy, the rise of open source (OS), the central role of standards in digital markets, and rapid advances in AI, there is a need for updated guidance on how technical collaborations should be evaluated under the antitrust laws. Here is an outline of the key points I made in my response. If You are interested, you can read my full comment at https://www.regulations.gov/comment/ATR-2026-0001-0018.
My response focuses on two primary areas of concern:
- Collaboration in OS-based standards development
- Policy trends favoring OS and open-weight (OW) AI models
Across both areas, the core issue is how evolving collaboration models affect competition, innovation, and the diversity of business models.
1. OS and Standards Development
- OS has evolved from a decentralized, community-driven model into a strategic tool for large firms to reduce costs, accelerate development, and support commercial offerings.
- While OS can promote innovation and competition, it may also reinforce market power, particularly in adjacent or downstream markets.
Shift in standards development:
- Traditional standards: Developed through SDOs using consensus processes and FRAND licensing.
- Emerging trend: OS implementations developed in OS communities or as OS projects increasingly shape or replace formal standards, sometimes becoming de facto standards.
Potential benefits:
- Faster and more efficient standards development
- Potentially broader SME participation
- Potential for improved technical outcomes
Key risks:
- OS communities often lack the same due-process safeguards found in SDOs, i.e., consensus-based voting, balance, etc.
- Disproportionate influence by a small number of well-resourced firms.
- Promoting market adoption of OS standards with misleading “royalty-free” claims despite third-party patent rights.
- Reduced long-term R&D incentives if innovations can be replicated without timely compensation.
Other Competitive implications:
- Reduced competition among independent implementations when standards are tied to a single OS implementation.
- Constraints on how firms commercialize technology by promoting OS-only business models.
2. AI Policy and OS/OW Models
- The term “open” in AI is often undefined, creating ambiguity (e.g., whether it implies fully unrestricted, royalty-free access).
- AI development requires significant investment, supported by diverse models (proprietary, hybrid, and open).
Concerns:
- Policies favoring fully open models may disadvantage IP-reliant firms
- Risk of weakened incentives for large-scale R&D
- Potential narrowing of business model diversity
Key Policy Considerations:
- Competition policy should consider not only competition among products, but also among business models.
- Intellectual property rights are essential to enabling diverse innovation strategies.
3. Transparency vs. Openness in AI
- Transparency in terms of explaining how an AI system operates and makes decisions is a key policy goal, but OS/OW AI may not effectively deliver it.
- Even developers often cannot fully explain AI system behavior.
- Favoring OS/OW AI systems may, in some cases, slow innovation especially in the long term by reducing business model competition.
Alternative approach:
- Standards-based testing and certification may better address AI risks while preserving competition.
4. Policy Recommendations My response suggests that the Agencies update the 2000 Collaboration Guidelines to reflect modern technology markets by:
- Addressing the competitive implications of OS-based standards and OS/OW AI collaborations
- Recognizing the importance of business model competition
- Clarifying the role of intellectual property in collaborative ecosystems
- Distinguishing between openness and procompetitive outcomes
