LONDON, (Reuters) - Britain's national broadcaster the BBC is drawing up plans to build and train its own artificial intelligence model using its text archives, the corporation said on Thursday.
A BBC executive told a parliamentary committee earlier this month the organisation was looking at generative AI as a tool to support its production process and assessing a possible unilateral or partnership arrangement on training a model.
Asked about a Financial Times report on Thursday that the BBC was planning to build its own AI models, a spokesperson confirmed the BBC was looking at developing a Large Language Model using text-based content.
The output of any such model would only be available to the BBC, the spokesperson said.
The Financial Times also reported, citing sources, that the broadcaster was holding talks with technology firms over selling access to its archives, which could then be used to train AI models.
In response to that, the BBC said: "The BBC has no agreement with any organisation to use its archive to train their large language models that power generative AI tools for commercial use.
"We are looking at a wide range of issues such as potential bias within such models and how the BBC either in partnership or unilaterally can address these."