MOZILLA CTO: AI SHOULD BE BUILT LIKE THE INTERNET

Raffi Krikorian, Chief Technology Officer at Mozilla, advocates for a fundamental shift in how artificial intelligence is developed and deployed. Rather than concentrating AI power in proprietary systems controlled by a handful of companies, Krikorian envisions an architecture mirroring the internet's distributed design. The current AI landscape shows a clear divide. Consumers gravitate toward closed-source applications like OpenAI's ChatGPT and Anthropic's Claude, which offer polished interfaces and powerful capabilities. However, enterprises are moving in a different direction. Companies increasingly choose open-source AI models they can run on their own infrastructure. This approach grants organizations greater control over their systems, allowing customization to specific business needs and data requirements. Unlike proprietary platforms, open models eliminate vendor lock-in and reduce dependency on external providers' pricing decisions and service availability. Krikorian's comparison to the internet's architecture highlights a critical distinction. The early web succeeded through open standards and distributed infrastructure, enabling widespread innovation and preventing monopolistic control. Applying this principle to AI would democratize access and foster competition among developers. The shift toward open models addresses practical concerns beyond control. Organizations handling sensitive data prefer keeping AI processing in-house rather than uploading information to third-party services. Customizable models enable businesses to fine-tune performance for their specific applications without relying on generic, one-size-fits-all solutions. Mozilla itself has invested in this vision, supporting open-source AI initiatives and promoting browser-based AI capabilities. The foundation's push reflects broader industry momentum, with companies like Meta releasing open models and others building deployment infrastructure for decentralized AI. Krikorian's perspective represents growing recognition that AI's long-term health depends on preventing concentration of power. While proprietary models dominate consumer attention, the enterprise shift toward open alternatives suggests the AI ecosystem is evolving toward the distributed model Krikorian advocates—one more resilient, competitive, and accessible than today's closed environment.