The BBC is drawing up plans to build and train its own AI model using its own text files, Britain’s public broadcaster said today.

A BBC executive told a parliamentary committee earlier this month that the public agency was looking at Genetic Artificial Intelligence as a tool to support the production process and evaluate a possible unilateral or joint deal to train a model.

Asked about a today’s publication of the Financial Times that the BBC plans to build its own AI models, a spokesperson confirmed that the BBC is looking at developing a Large Language Model using text-based content.

Production of any such model will only be available to the BBC, the spokesman said.

The Financial Times also reported, citing sources, that the public body was in talks with technology companies to sell access to its files, which could then be used to train Artificial Intelligence models.

In response the BBC said: “The BBC has not entered into any agreement with any company to use its archive. Therefore, no company will be able to train with it the Big Model Languages ​​that feed AI production tools for commercial use.”

“We are looking at a wide range of issues such as potential bias in such models and how the BBC either collaboratively or unilaterally can address them,” he added.