Does an AI training licence eliminate every legal risk?
No: its value depends on the rights actually controlled and the uses granted. Copyright ownership, personal-data handling and restricted data transactions require separate checks; the DOJ programme illustrates a non-copyright constraint.
Is the Copyright Office training report a new law?
No; it is a policy and legal-analysis report. The Office’s study page still identifies Part 3 as a pre-publication version, so cite it with that status.
Does a fair-use win mean copying any archive is lawful?
No: §107 requires analysis of the particular use and relevant factors. Ask counsel to compare the actual collection, product and market facts rather than a headline.
Should we budget Bartz settlement payments as licensing income?
Keep litigation recovery separate from newly negotiated licence revenue. The administrator describes conditional distributions for qualifying claims, not an open licensing price list.
Are we automatically subject to California DROP because we licence articles?
No; first determine whether the legal definition of data broker applies to the entity and activity. CPPA guidance describes the duties of qualifying brokers, not all publishers.
Do EU model-provider duties become our duties as a US archive owner?
Not merely because the archive is licensed: identify each party’s actual role. Article 53 addresses GPAI providers, but buyer compliance can create contractual requests for provenance and documentation.
Does a content credential prove AI licensing authority?
No; the C2PA specification concerns integrity and provenance assertions. Match those assertions to contributor contracts and other rights evidence before relying on them.