Most organisations have a crisis communications plan – but few have tested whether it still works in a world where AI, not a journalist, is often the gatekeeper of your story. Jack Sice explores this in our latest blog.
For a long time, the crisis communications plan lived as a PDF on a shared drive. It was dusted off when something went wrong – at a time when crises moved at the speed of the news cycle. Facts got established, and a well-briefed spokesperson could set the record straight.
Social media came along and broke that model. As information, and misinformation, spread faster than any press office could respond, communications teams were forced to adapt to become faster, more agile and multi-channel in approach.
Now, we’re in the midst of another monumental shift. Artificial intelligence has stepped into the middle of the information chain – summarising, aggregating and drawing conclusions from whatever signals it can find.
How stakeholders learn about your organisation has changed radically in a relatively short space of time. Increasingly, they meet you first through an AI-generated summary, which has been shaped by years of accumulated content, comprising news coverage, social posts and other sources that have become more authoritative in their own right, such as Reddit threads.
AI tools take all of that and compress it into one confident-sounding answer. That has three consequences worth taking seriously:
Another important point to consider is that AI doesn’t forget. Though AI models do tend to have a ‘recency bias’ that favours newer content, a historic incident can surface just as readily as one from last week if you ask a model about your organisation. A well-known disaster surfaced almost immediately from a quick search about an oil major’s environmental efforts, despite the incident taking place in the same year as Instagram was founded – more than 15 years ago.
Crises that once might have run their course and faded, banished to the back page on Google Search, now have the potential to trail a brand, organisation or business long after the original story has been forgotten by everyone – except the model.
Most crisis plans were, clearly, never built with that in mind.
As a first step, organisations need to know what AI tools are already saying about them. That means auditing current outputs, correcting inaccuracies at source and making sure owned channels are publishing content that is clear and factual enough to shape what the models pick up.
It should be thought of as ongoing ‘reputational maintenance’. AI is always building a picture of you, so the groundwork has to be continuous, not an occasional box-ticking exercise. This is where PR has an important role to play, as ensuring that your organisation is issuing a steady drumbeat of news and content can help lay a positive and accurate foundation in the event of a crisis, particularly given AI models strongly favour earned media sources – our bread and butter.
During a crisis, statements need to reach the channels AI actually draws on, not just the traditional media list. That’s not to say that getting the message out to the titles on that list isn’t critical – the BBC and The New York Times will continue to be among the most authoritative sources. What it means is that the multi-channel mindset, and consistency in messaging across these, has never been more important.
After the story has moved on for everyone else, it may not have moved on for AI. The information environment needs monitoring closely and precisely, as inaccuracies can sit in AI responses long after the news cycle has ended, reading as fact to anyone who asks.
The organisations handling crisis communications well aren’t just the fastest responders.
Although speed is still integral to determining how well an organisation deals with an unfolding crisis, those that successfully navigate these will be the ones building AI visibility into crisis planning before an incident happens, rather than scrambling to manage it once the story is already circulating.
In practice, that comes down to three things: auditing what AI tools currently say about the organisation and fixing what’s wrong at source; distributing statements through the channels AI draws on; and monitoring the information environment on an ongoing basis – not just when something goes wrong.
Crisis communications has always rewarded preparation over reaction. In the age of AI, the gap between organisations that have prepared and those that haven’t is wider than it has ever been.
Want to learn more? Find out about Signal, firstlight’s GEO offer, here.