Issue #2: Foundation moves, sustainability pressure, and AI governance friction

This week in Open Source Funded, the clearest pattern is institutionalization.

Several projects moved under foundation governance or advanced inside existing foundation maturity ladders. Around that, the rest of the week’s stories kept returning to the same questions: who is paying for the commons, what kinds of support actually matter, and how much extra policy and review work AI is now creating for open source projects.

Per this issue’s editorial rules, this section also includes projects advancing to a new stage inside an existing foundation.

This is exactly the kind of institutional movement worth tracking. It is not the same as direct project funding, but foundation governance often brings clearer stewardship, more neutral branding, stronger IP handling, and better odds of long-term ecosystem participation.

The most directly relevant funding story in this week’s set is the GNOME Foundation’s new fellowship program. That matters because it is explicitly about funding contributor work aimed at long-term sustainability, not just paying for infrastructure or event operations.

At the ecosystem level, OpenSSF’s March update also stood out because it highlighted $12.5 million in new funding from Anthropic, AWS, Google, and others for open source and AI security work. That is still top-down money, but it is top-down money aimed at shared public goods.

Sources: Introducing the GNOME Fellowship program, OpenSSF Newsletter – March 2026

A lot of this week’s support news was not one giant grant. Instead, it was a series of smaller institutional commitments that still matter when viewed together:

It is easy to treat this kind of announcement as routine foundation housekeeping. But these are still useful signals that open source institutions continue to attract budget, legitimacy, and organizational attention.

LibreOffice is adding a donation banner to its Start Center, and the reaction is more interesting than the UI change itself.

The banner is modest. The revealing part is how quickly even a small, visible funding prompt can trigger backlash. It is a recurring contradiction in open source: many people agree maintainers and institutions need more support right up until a project asks for it in public.

Source: LibreOffice is Adding a Donation Banner That Might Annoy Some Users

This week also brought two strong statements of the underlying sustainability problem.

The Register’s opinion piece argued that open source cannot keep behaving like a tip jar and that maintainers need to charge for access or otherwise stop depending on charity from billion-dollar users. FOSS Force’s essay made a related complaint from a different angle, arguing that cloud companies continue to extract enormous value from open source while underfunding the commons that make that value possible.

They are opinion pieces, not neutral reporting, but they fit the week well. Between the LibreOffice donation-banner debate and the steady stream of foundation support announcements, the same unresolved tension keeps showing up: open source creates huge economic value, but there is still no settled agreement on how that value should flow back upstream.

Sources: Open source isn’t a tip jar – it’s time to charge for access, Cloud Czars Treat Open Source Like They Do California

GitHub says Copilot interaction data will be used for model training by default unless users opt out. That turns ordinary tool use into a governance question about whether prompts, code snippets, and surrounding context should be repurposed as training input.

At the project level, Gedit made a different kind of move by saying it will not accept AI- or LLM-generated contributions. That matters because these debates stop being abstract as soon as they become contribution-policy rules.

Sources: Updates to GitHub Copilot interaction data usage policy, Gedit Aims For More Frequent Releases, Bans AI / LLM Contributions

The New Stack reports that maintainers are increasingly dealing with low-quality AI-generated pull requests and contribution noise. That is one of the most practical AI stories in open source right now: even when generated changes are not malicious, they still create triage and review costs.

At the same time, Solo.io launched the open source agentevals project and said it donated its agentregistry project to CNCF. That shows the other side of the same trend: the ecosystem is not just resisting AI pressure, it is also building new open source infrastructure around agent evaluation and governance.

And outside foundation governance, Astral’s acquisition by OpenAI is another reminder that AI companies are steadily absorbing more of the developer tooling ecosystem.

Sources: 96% of codebases rely on open source, and AI slop is putting them at risk, Solo.io launches agentevals to solve agentic AI’s “biggest unsolved problem”, Astral has been acquired by OpenAI

We re-checked the links in jobs.yaml before publishing. The listings below all still resolved to live job or application pages at publication time.