AI Data Broker for Film & Video Production: Film Archive Licensing

If you run a production company, this probably sounds familiar: ten or twenty years of raw footage, B-roll, and outtakes spread across hard drives and LTO tapes. Almost none of it was published, and it earns nothing.

At the same time, you keep reading that AI companies are scraping video from anywhere they can find it, and that some are paying for it. So the questions are fair. Can we get paid for our archive? Is any of it ours to sell? And how do we avoid signing away something we will regret?

They are the right questions. An archive's value depends less on how good the footage looks than on whether you can prove you own it, show who is in it, and hand over a clean, documented package.

The main buyers are companies building generative video models, and the large AI labs adding video to their models. They need footage they can defend in court, which is why licensed archives have become valuable as scraping has become legally risky. According to Bloomberg, AI companies have paid creators between US$1 and US$4 per minute for unpublished video, with ordinary clips at the low end. Individual deals were often worth thousands of dollars.

Studios are also licensing at the top end. In September 2024, Runway announced a partnership with Lionsgate to create and train a custom AI model on Lionsgate's catalogue, a library of more than 20,000 film and television titles. It shows the direction: rights holders licensing on their own terms rather than being scraped.

Volume and uniqueness drive value. Buyers think in hundreds or thousands of hours, so a smaller archive is often worth more when it is packaged around a clear niche, such as aerial work, a region, or a specialist subject.

British Columbia is one of the largest production centres in the country. CMPA's Profile 2025 puts total film and television production in B.C. at $2.69 billion in 2024/25, level with Quebec and behind only Ontario. About $2.1 billion of that was foreign location and service production.

That matters for licensing. On service productions, the footage generally belongs to the studio that commissioned the work, under contract, not to the local company that shot it. The archives that B.C. and Canadian companies can actually license tend to sit with Canadian-owned producers, documentary makers, commercial and corporate video houses, drone operators, and stock shooters.

The legal backdrop also favours licensing over scraping. Canada has no text and data mining exception in its Copyright Act, and copyright collectives have asked Ottawa not to add one. The federal AI strategy released in June 2026 did not address copyright, while an April 2026 House of Commons report recommended stronger consent and licensing requirements for training on copyrighted work. For a producer, a clean, documented licence is the safest way to turn an archive into revenue, and the most credible form of data monetization in this industry.