The Future of News in an AI Era – Combatting Fake News

When Apple announced their Apple Intelligence software it heralded a new era of summarised information from a huge volume of sources into a nice easily digestible format. Except it didn’t. In January of 2025 they stripped back the service to no longer include any news summary due to major instances of fake news being displayed to the reader. This was followed by a report in February 2025 a news report was published by the BBC claiming 51% of AI news generated summaries contain “significant issues of some form”. Following this, we’ve explored where is AI facing roadblocks to providing the service expected, and how  Digital PR teams can support brands to navigate these.

What is the AI state of play?

Even market leaders OpenAI state that their o-1 preview model will produce factual errors 50% of the time¹, and this grows to a shocking 90% for some of their smaller less powerful offerings. This means anyone seeking to use the chatbots in the same way they would a search engine will be very disappointed with the outcome. 

In order to understand the growing problem, it is important to put into context how generative AI algorithms work. Using probabilities they produce the average next word based on the previous, an inherent complexity sometimes not fully understood by anyone on the outside of the companies involved . The model has a logic that can take an input and summarise a huge corpus of training information into what the average response to a question would be. It is akin to writing down the average answer in a pub quiz, a great strategy only if everyone else has already written down the right answer.

It has been observed that these models can be capitalised on for online visibility, at least currently. Commonly associating your brand with certain terms will often directly mean that an LLM will refer to your brand with those terms. This gives Digital PR even more power than normal when it comes to improving brand visibility.

Links play a key part SEO’s strategies when it comes to Google algorithms, follow/nofollow, syndications and relevance are all quoted as being of varying importance depending on your school of thought. It seems apparent to all involved in the industry that LLM’s shift that narrative, where just a brand mention alone and the context around it holds greater weight. Ensuring brands are mentioned alongside relevant information will ensure they appear more frequently in responses both within chatbots and in whatever form “AI search” takes moving forward.

Why is this important for Digital PR and the news cycle?

As with many other industries the influx of AI generated content into the marketplace has caused huge challenges for Digital PR as an industry. In much the same way LinkedIn guru’s offer of 1,000 articles a day forced Google into many algorithm updates, those instances of unscrupulous PR teams offering fake experts via HARO forced the platform to no longer be a viable place for journalists to find experts. Though recent activity suggest the company believed they have found a solution and are preparing to reactivate the brand. 

Creating large volumes of average content has never been easier and journalists’ inboxes are stacking up with ChatGPT derived top tips and expert insight. If the BBC statistics are even close to accurate, that is a worrying level of fake news being pitched in daily at a huge scale. The only thing certain is as Digital PR’s, it’s never been more important to deliver a quality service to both clients and the journalists we engage.

So how can Digital PR ensure it does not fall foul of AI?

Understand how generative AI works

The key to avoiding average and inaccurate content is understanding that by design a large language model will produce the most probabilistic response. This by its very nature is the most average response possible, regardless of any sense of accuracy. Using it for creating PR content won’t help your story stand out, and will only serve as a barrier between PR’s and journalists.

Using unique data

Though large hero campaigns have become less popular in recent years. Well angled, well researched stories are still popular with journalists. Providing unique data is something that AI cannot do and helps to create a new and innovative story for a journalist to cover, rather than just re-hashing old ideas it also positions clients as experts and thought leaders in their industry. 

Use client expertise

Connecting journalists with experts has long since been a crucial part of Digital PR. The fall of HARO highlighted how easy it is to produce fake quotes and send them to journalists en masse. The easiest way to counteract this trend is to form relationships with journalists, introduce them to your experts and build trust so that your comments can be covered. 

1. https://openai.com/index/introducing-simpleqa/