Creative Commons is moving into one of the most contested questions in AI: how openly shared data should be used by machines. The nonprofit announced CC signals on Wednesday, describing it as a way for dataset holders to spell out how their content can or cannot be reused, including for training AI models.
The project is still early, but its goal is clear. Creative Commons wants a framework that can support openness online while responding to the growing demand for data that powers AI systems.
Why CC signals matters now
Creative Commons helped popularize licensing that lets creators share their work while still retaining copyright. That history matters because the organization is now trying to apply a similar kind of thinking to machine reuse.
The tension is not only about whether AI companies need data. It is also about what happens to the open internet if data extraction continues without clearer expectations. As Creative Commons explains in a blog post, ongoing extraction could push entities to wall off their sites or put data behind paywalls instead of sharing access.
CC signals is meant to offer another path. Rather than forcing dataset holders to choose only between total openness and tighter restrictions, the project aims to give them a way to communicate reuse terms for machines.
What the framework is trying to do
At its core, CC signals is a legal and technical proposal for dataset sharing. It is meant to operate between those who control data and those who use data to train AI.
The project would let dataset holders give more detail about how their content may be reused. That includes cases where the reuse is connected to training AI models.
Creative Commons describes the work as a set of tools with a range of legal enforceability and ethical weight. The comparison is intentional: the organization points to CC licenses, which today cover billions of openly licensed creative works online.
“CC signals are designed to sustain the commons in the age of AI,” said Anna Tumadóttir, Creative Commons CEO, in an announcement. “Just as the CC licenses helped build the open web, we believe CC signals will help shape an open AI ecosystem grounded in reciprocity.”
The word reciprocity is central to the idea. The project is not described as a simple blocking tool. It is framed as a way to keep shared resources available while giving the people and organizations that control datasets a clearer voice in how those resources are used.
The pressure around AI training data
Demand for a tool like this is increasing as companies reconsider their policies and terms of service. Some are trying to limit AI training on their data. Others are explaining how user data may be used for purposes related to AI.
The source article points to several examples of how fragmented the current response has become:
- X initially made a change that allowed third parties to train their models on its public data, then later reversed that.
- Reddit is using its robots.txt file, which is meant to tell automated web crawlers whether they can access its site, to restrict bots from scraping its data for training AI.
- Cloudflare is looking toward a solution that would charge AI bots for scraping, as well as tools for confusing them.
- Open source developers have built tools to slow down and waste the resources of AI crawlers that did not respect their “no crawl” directives.
Those approaches show the same underlying concern from different angles. Dataset holders and online platforms want more control over machine access. At the same time, the open web depends on access, reuse and sharing.
CC signals enters that debate as a framework rather than a single defensive tactic. Its promise is that clearer signals could reduce the need for more closed-off responses, while still giving data controllers a way to set expectations.
What happens next
The project is only beginning to take shape. Early designs have been published on the CC website and GitHub page.
Creative Commons is also asking for public feedback before its planned alpha launch, described as an early test, in November 2025. The organization will host a series of town halls for feedback and questions.
That feedback stage matters because CC signals is meant to sit between different groups with different interests. Dataset holders need usable ways to express their terms. AI developers need signals that can be understood and acted on. The broader internet depends on rules and norms that do not push useful data into closed spaces.
The project does not claim to settle every dispute around AI training. But it does identify a practical gap: the lack of a shared framework for communicating machine reuse permissions in a way that carries both legal and ethical force.
If CC signals succeeds, it could become part of the infrastructure for a more open AI ecosystem. If it fails to gain adoption, the pressure toward paywalls, bot restrictions and anti-scraping tools may keep growing. For Creative Commons, the bet is that the commons can survive the AI era only if sharing comes with clearer terms.