LONDON — Over the years, radio has leveraged new technologies to enhance its news coverage, bringing listeners the latest stories as they unfold. In the digital age, computer-based systems, notably the internet, have expanded this and now AI is set to change news production further still.
Somewhat ironically, despite being the direct result of a new technology, radio did not immediately capitalize on that potential in the realm of news. In its early days, the BBC produced news bulletins based on copy from wire agencies well into the 1930s. This was partly due to opposition from newspapers, which recognized the threat a BBC with its own newsroom could pose.
The corporation did employ journalists, but primarily for preparing newscasts and fact-checking. In 1936, one of them wrote a memo to the chief news editor suggesting that bulletins should incorporate eyewitness accounts of an event and “a short talk… on some topic of the day.” This was the 22-year-old Richard Dimbleby, who would go on to become the BBC’s first official war correspondent and a pioneering figure in broadcast journalism in the United Kingdom. He put his suggestions into practice later that year when the Crystal Palace caught fire and he broadcast live from the scene over a telephone line.
From that point, radio broadcasters began developing their own news operations, providing near-instantaneous reporting, often with all-important actuality. Technology made this possible, with the emergence of dedicated copper circuits for contributions — later superseded by ISDN and then IP — and, post-war, quarter-inch tape recorders, which allowed journalists to edit their reports into tight packages. The introduction of computer-based newsroom systems in the 1980s streamlined the news production process. These have continued to evolve to the point where AI, particularly in its generative form, is now seen by broadcasters as the way forward in creating efficient and cost-effective news output.
Benefits and risks
According to “The Future of Newsroom Workflows” report published earlier this year by Caretta Research on behalf of Dalet, nearly 80% of news organizations see AI in the newsroom as “a benefit.” A small proportion of those polled (14%) expressed concerns about AI threatening jobs and creating “new risks.” The report quotes the head of technology at a local news station as saying that AI should be utilized to “make operations more efficient without compromising our integrity,” emphasizing that people should always be involved in the process.
Dalet highlights four areas where AI can bring benefits in news production: Automating regular jobs, including tagging, formatting and producing metadata; captioning and categorizing live and recorded material; translating and versioning; summarizing stories and suggesting story angles; and personalization, such as selecting stories and clips based on listener preferences.
The BBC announced in June that it is now conducting public tests of two pilot schemes involving GenAI — one automating mundane tasks and the other employing summarization.
Style Assist converts stories received from the Local Democracy Reporting Service, a public service news agency funded by the BBC, into the broadcaster’s house style. Based on a large language model, the system has been trained on thousands of BBC-written features and automatically produces a revised draft of an original piece with the relevant style and tone. A journalist then reviews the story to ensure it is clear and accurate before approving it for publication.
The At a Glance summaries app produces shortened versions of longer news features. Rhodri Talfan Davies, the BBC’s executive sponsor of GenAI, explains that the aim is to produce “short, scannable bullet-point summaries,” which he says have proved popular with readers of the BBC News website; especially younger ones. As with Style Assist, the summaries are reviewed and edited before being published. While neither app is audio-based or explicitly aimed at radio, they will play a part in the multimedia approach that stations are taking today.
The ultimate implementation of AI in news is fully automated and virtualized newscasts.
Increasing efficiencies, upsetting unions
The 2025 EBU News Report, subtitled “Leading Newsroom in the Age of GenAI,” identifies back-end automation (tagging, transcription and copy editing) as the most important use case for AI in news this year, according to 60% of people surveyed. This is followed by distribution, personalization and recommendations/alerts, as well as content creation with human oversight (summaries, headlines, video and audio). Newsgathering, identifying stories and studying data come in last, with only 24% seeing it as a critical application.
The report features nine case studies of broadcasters utilizing AI in various ways. Most relevant to radio is Swedish Radio’s audio-focused text generator, produced by the in-house AI team, which creates headlines and three bullet-point summaries for each of the 370 sound clips the broadcaster produces daily. Transcripts are fed into the OpenAI API, providing suggestions for the copy to the human editor. SR found that these were of “good quality” and accurate, resulting in increased efficiency.
Low awareness
Despite such advances, awareness in some parts of the industry appears to be low. A survey published this year by the Reuters Institute and the University of Oxford found that only 7% of U.K. journalists “were aware of working in a newsroom that used automated text production.” The potential for AI in news production is considerable, but while broadcasters have adopted some aspects of it, these are still in their early stages.
The ultimate implementation of AI in news is fully automated and virtualized newscasts. This is happening, but on a relatively small scale, with broadcasters understandably wary of upsetting journalists’ unions by implying that they are phasing out their members. Among those flirting with AI bulletins is U.K. platform GB News Radio, which last year began broadcasting 30-second AI-generated sports updates featuring the cloned voice of newsreader Tatiana Sanchez.
The technology to automatically produce the news is now widely available. Aiir offers the Audio.co system (a rebrand of RadioNews AI), which can generate AI-assisted news programs. It sources stories from selected websites and RSS feeds and allows the user to choose an AI voice to deliver them. While the technology is impressive, if a little troubling, it does seem like a return to the days before Dimbleby wrote his famous memo.
The author trained as a radio journalist and worked for British Forces Broadcasting Services Radio as a technical operator, producer and presenter before moving into magazine writing during the late 1980s. He recently returned to radio through his involvement in an online station where he lives on the south coast of England.
This article originally appeared in the RedTech International special publication Radio Futures 2025
These stories might interest you
Radio Futures explores AI’s next wave
