Data access is central for the development of Artificial Intelligence (AI). Data may be however subject to a multitude of rights and actors (e.g. copyrighted works or personal data posted on social media or collected through captcha and used for AI training purposes), leading to a fragmented legal framework and legal frictions. This contribution explores how to apply open and free licenses to all kind of data and collective governance of data through data trusts, in order to allow complex management of data subject to multiple legal regimes and to increase data flows, while improving trust and control over the data. Free or open licenses are traditionally used in connection with Open Source and Creative commons, but could be eventually extended to personal data due to their common characteristics. Collective governance could be improved by creating data trusts, inspired by pilot projects and by the collective management of copyright.