What does O'Reilly mean by open-source AI beyond just open-weight models?

He means unlocking the whole stack of an AI system—the model, the harness, and the application—with a clean separation between them, giving control to designers and users rather than building what he calls an "architecture of control."

Why does O'Reilly think frontier AI models like Claude are the wrong direction?

He argues that big models are optimizing for particular use cases the labs have chosen, not necessarily the ones ordinary people want, and that they're actually pushing further away from what society needs. He points out that people are noting lower-level models write better than newer frontier models like Fable and Sol.

How does O'Reilly use AI in his own work?

He uses AI for brainstorming and functional writing, such as turning hour-long interviews into usable drafts, but says he does not use it to write the main editorial content—that remains human-written.

Large Language ModelsOpen-Source AIWIRED AIAug 14, 2026Aug 15, 2026, 01:00 JST

O'Reilly: Big AI Labs Building 'Control,' Not Freedom

Send on LINE

3 Key Points

What happened
Tim O'Reilly, the publisher and tech pioneer, is promoting open-source AI—not just open-weight models, but the entire stack—to give users and designers control over AI systems rather than locking them into proprietary products from hyperscalers like OpenAI and Anthropic.
Why it matters
O'Reilly argues that today's frontier AI models are optimizing for use cases big labs want, not what ordinary people need, and that the real innovation will happen in widely distributed lower-level models, not in the few companies dominating the frontier. He sees the concentration of AI capital as a failed venture-capital model that chokes out alternative paths, much like Uber and Lyft did in transportation.
What to watch
O'Reilly is working on an open-memory consortium through his nonprofit, the AI Disclosures Project, as a counterweight to Mark Zuckerberg's vision of locking users into Meta's AI. He predicts frontier models may become like mainframes—useful for hard problems but not what gets diffused through society.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

O'Reilly's critique centers on a fundamental mismatch between what the major AI labs are building and what markets actually need. He argues that because a handful of companies control the capital and resources for frontier AI, they have the power to shape the technology in ways that reflect their business interests—keeping users locked in—rather than what customers want. This echoes his longstanding philosophy that value creation should be measured by what you give away, not what you capture.

He draws a parallel to the venture-capital model that took over Silicon Valley after 2010, when firms like Uber and Lyft used massive funding to subsidize their way to dominance, effectively choosing winners rather than letting markets do so. In AI, he sees the same pattern: all the capital is funneling into a few companies pursuing frontier models, which he believes is a strategically flawed bet. His comparison to the 1990s PC wars is telling—while everyone focused on who would dominate personal computers, the web emerged from outside that battle and ultimately changed everything. O'Reilly suggests open-source AI, supported not by venture capital but by communities and smaller innovators, will be that unexpected disruption.

FAQ

What does O'Reilly mean by open-source AI beyond just open-weight models?: He means unlocking the whole stack of an AI system—the model, the harness, and the application—with a clean separation between them, giving control to designers and users rather than building what he calls an "architecture of control."
Why does O'Reilly think frontier AI models like Claude are the wrong direction?: He argues that big models are optimizing for particular use cases the labs have chosen, not necessarily the ones ordinary people want, and that they're actually pushing further away from what society needs. He points out that people are noting lower-level models write better than newer frontier models like Fable and Sol.
How does O'Reilly use AI in his own work?: He uses AI for brainstorming and functional writing, such as turning hour-long interviews into usable drafts, but says he does not use it to write the main editorial content—that remains human-written.

WIRED AIRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.