What is the relationship between journalists and artificial intelligence in 2025? AI has been used in newsrooms for several years now, but the relationship between the human and the tech is still in flux. The opening panel at the JournalismAI Festival delved into how three big news organisations (Financial Times, BBC, and The Guardian) are using it. Is AI a useful assistant for journalists, or a co-writer?

Check out all of the key takeaways from the conference with our overview.

AI experimentation and assistance

A lot of the newsrooms who have implemented AI tools are already seeing gains, and for Liz John, product director at the Financial Times, that has been with comments on articles:

‘We’ve used AI to actually drive people more into the comments. The audience engagement team reviews a question that is generated by an LLM based on the article and is trying to spark a discussion. That question is then posted onto the article, and then takes you to the comments. It’s increased comment rate by 3.5%.’

At the BBC, there has been a lot of experimenting going on with Natalie Malinarich, director of growth, innovation and AI, talking about several interesting pilots including one in the sport team, turning text into short podcasts, and vice versa. One pilot is now also set to be integrated properly to offer greater assistance, as Natalie explained:

‘We have had a very successful pilot that’s about to be integrated into our CMS and rolled out more widely, which is around our local democracy reporting. There are 150 reporters across the country and they don’t file in BBC style. Using a BBC LLM, we’ve given them the ability to pipe straight into the CMS, and if the LLM does the first pass and puts it into BBC style and comes up with some suggestions or flags up some issues. That’s been really successful, and the accuracy rates are quite high.’

The changing newsroom

The invention of AI was always going to have an impact on the way the newsroom worked and at The Guardian, the major change has been Google Gemini being approved internally for everyone. Chris Moran, head of editorial innovation at The Guardian, stressed that using AI is ‘a personal decision about where it fits into your natural workflow’ as a journalist. However, he does feel that the ‘integrations are proving to people who have previously been worried about it, that there are things that can help them’.

For Liz John and the Financial Times, the change seems to be having even more of an impact:

‘It is shifting much more to the writer right now, because it is so much more embedded in the editor, rather than at a separate step where someone goes and checks that. So already there is shifting workload and shifting roles between the newsroom jobs.’

These changes need to be thought out and Liz John spoke about the COM model – Capability, Opportunity and Motivations – as a way to handle this:

‘In terms of capability, are we offering the tools for people to behave differently? Then, are we giving them the opportunity, meaning time, for instance, or even coaching? And then the motivation, I guess, is the critical one, and that’s what we were struggling with a couple of years ago, when people were scared that if they engaged, they were going to work themselves out of their jobs. So what is the positive thing? What are they working towards? What are they creating? Figuring that out, I think, might be more important for the system to be self sustainable.’

Being clear on where AI has been used

With AI being used more often either during the process of newsgathering or helping to actually put the article together, publishers need to decide whether to disclose this, and when. For the Financial Times readers, Liz John said that an AI disclaimer currently ‘erodes trust’ and gives the feeling of it being ‘cheap’.

An organisation like the BBC prides itself on its impartiality and being as open and honest with its audience as possible, and was a point that Natalie Malinarich touched upon:

‘We try to be very descriptive about what we do, so it’s very clear which bits AI has assisted with. We do disclose everything, so whether it’s assistance with translation, assistance with summarisation, it is all disclosed. But I also think it will change over time, as people become used to the idea of AI.’

The Guardian have had their editorial code visible to the public for a number of years now and Chris Moran doesn’t believe that it needs to be too complicated when it comes to AI either: 

’Like a huge amount of existing journalistic practice, ethics and standards should moderate sensibly this new technology’.