When OpenAI announced it was co-founding the Agentic AI Foundation under the Linux Foundation, you’d think I’d be impressed. After all, the group includes Anthropic, Block, Google, Microsoft, AWS, Bloomberg, and Cloudflare. That is a who’s who of AI power brokers. But instead of feeling optimistic, I found myself wondering why all of these companies suddenly care so much about “open” agentic AI after spending years pushing tightly controlled, closed systems. The timing feels a little too convenient, and frankly, a bit like smoke and mirrors.
OpenAI’s headline contribution is AGENTS.md, a simple Markdown file that tells agents how to behave inside a project. It has already spread across tens of thousands of repositories, so donating it costs them nothing. In fact, it cements their preferred structure as the so-called neutral standard. If that looks like generosity, look twice. It is more like planting a flag before anyone else can. When you give away something low-risk, you get the PR win without surrendering any real leverage.
The whole pitch leans heavily on the idea that agentic AI is moving into real-world production and needs guidance. Fair. But if these companies truly believed in openness, they would open the things that matter: their models, their training data, their safety tooling, and their decision-making processes. Instead, they are offering a governance shell that feels designed to calm developers who are increasingly tired of black-box AI running the show. It is hard not to view the foundation as a pre-emptive shield against criticism rather than a genuine invitation to collaborate.
There is also the regulatory angle. With governments starting to pressure AI companies about transparency and interoperability, this foundation looks like a soft attempt at self-regulation. It lets the industry say “Don’t worry, we’ve created a neutral body” while keeping the same companies steering the narrative behind the scenes. The Linux Foundation involvement adds credibility, but that doesn’t erase the influence of the corporations funding the entire effort. A directed fund is not the same as a grassroots project.
Anthropic is donating MCP, Block is donating goose, and each partner is tossing in something meant to look like skin in the game. Yet none of these contributions touch the core assets that actually define their competitive edge. The open-source world knows this pattern well. When tech giants want community goodwill, they open small tools around the edges while keeping the real crown jewels locked away.
OpenAI emphasizes that the Agentic AI Foundation will shape a shared, neutral platform for building agentic systems. Maybe it will. But neutrality is hard to believe when every major player at the table stands to benefit from guiding the standards that everyone else must follow. When a handful of companies write the rules, “open” starts to feel like a marketing term instead of a commitment.
The foundation is inviting developers, researchers, and enterprises to join. I’m sure many will. But before we celebrate it as progress, it is worth asking the blunt question. Is this really openness, or just a polished way to maintain control while appearing cooperative? From where I’m standing, it looks more like the industry trying to manage optics rather than surrender even an inch of influence.
Support independent tech journalism
NERDS.xyz is independently owned and operated. If you enjoy my coverage of Linux, AI, hardware, cybersecurity, and tech culture, consider supporting the site on Ko-fi.
Support NERDS.xyz