Mozilla’s CTO thinks AI should be built like the internet

Mozilla's Chief Technology Officer, Raffi Krikorian, argues that artificial intelligence should be developed along the principles of openness that have long defined the internet. In a recent interview with Rest of World, Krikorian emphasized the importance of open-source and open-weight AI models as viable alternatives to proprietary systems such as Anthropic’s Claude and OpenAI’s GPT series. His comments come amid a growing movement advocating for open AI, driven by both technological progress and increasing concerns around data control and transparency. Krikorian pointed to a recent report by Mozilla, which noted that the performance gap between leading open-source AI models and proprietary ones has shrunk to just 3 percent. This narrowing of capabilities suggests that open-source models are becoming increasingly competitive with closed systems. The report also highlighted the rapid adoption of open models, citing that in February 2026, Alibaba’s open-source model Qwen was downloaded more times than the next eight models combined. This trend has been further accelerated by Meta’s release of a new open-weight model on August 10, with promises of additional releases to follow. According to Krikorian, the shift towards open AI is not merely a technical evolution but a necessary step for businesses and governments seeking greater autonomy over their data and infrastructure. He drew parallels between the rise of open-source AI and the early days of Linux, which began as a small project but eventually became foundational to modern computing through its adoption by major corporations like Google. Just as Linux evolved into the backbone of smartphones and servers worldwide, Krikorian sees open-source AI gaining traction in enterprise environments due to cost-effectiveness, customization, and compliance requirements. Despite the growing influence of open-source AI, Krikorian acknowledged that public awareness of these developments lags behind. While individuals often rely on proprietary tools like ChatGPT and Claude, corporate IT and HR departments are increasingly turning to open models for internal workflows. This shift is driven by factors such as cost efficiency, the need for data sovereignty in regulated industries, and the desire to maintain operations within secure firewalls. However, Krikorian expressed confusion over why more companies are not publicly celebrating the financial benefits of adopting open-source solutions. He also raised questions about potential barriers to wider recognition of open-source AI, suggesting that geopolitical tensions might contribute to hesitancy. Many prominent open-source models originate from China, and while there is no definitive evidence of systemic distrust, Krikorian implied that such perceptions could be influencing decisions made by Western enterprises. Nevertheless, he stressed that the complexity of understanding and deploying open-source AI models may be a more immediate obstacle than political considerations. Krikorian’s perspective aligns with broader industry trends showing increased experimentation with open and open-weight AI models. Companies are recognizing the strategic advantages of having access to customizable, transparent, and less costly AI solutions. As the landscape continues to evolve, the role of open-source AI in shaping future technologies and business practices is likely to expand significantly. Mozilla’s advocacy for open AI reflects its longstanding commitment to promoting an accessible and inclusive digital environment. With the release of its latest report and ongoing discussions led by figures like Krikorian, the push for open AI appears poised to gain even more momentum in the coming months.