Friday 20 November 2026

The UK’s digital industry, overheard.

Get The Natter

Home / Opinion

Opinion

Crisis plans must catch up with AI-driven misinformation, argues TMW’s Fiona Flanagan

Deepfakes and synthetic news have cut the time communications teams have to respond, and the first hour now decides whether a crisis stays manageable, she says.

Illustrative image accompanying an opinion piece on AI and crisis communications

Fiona Flanagan of TMW, which is part of Accenture Song, argues that generative AI has removed the breathing space communications teams once relied on when a crisis broke, and that many organisations’ playbooks have not adjusted.

In the past, she says, a faulty product, a cyber-attack or a senior executive in the headlines for the wrong reasons still left teams time to establish the facts, get stakeholders aligned and shape a reply before the story ran away from them. In her view that assumption “no longer holds”.

Generative AI has shrunk the gap between making content and spreading it. Fake material can be produced in seconds, copied at scale and amplified worldwide before most organisations have even checked what they are looking at.

Invoking the old line about a lie circling the globe before the truth has got its shoes on, she writes: “With the speed of things now, the truth won’t even have the chance to pick which pair of shoes to put on before the lie is out there.”

The task, she argues, is no longer only managing a crisis but “identifying reality fast enough to respond”.

Flanagan calls deepfakes, synthetic news and AI-generated impersonation boardroom issues. Fraud, reputational harm and lost stakeholder trust increasingly stem from deception enabled by AI, she says, yet many organisations are unprepared to respond at the speed required.

She cites figures putting the global number of deepfakes at 500,000 in 2023 and over eight million two years later, a 1,500% rise in two years.

Accenture’s State of Cybersecurity Resilience 2025 report, she adds, found ninety percent of organisations worldwide cannot defend themselves against AI-driven threats, while nearly two-thirds lack both a cohesive strategy and the technical means to respond.

The first hour

Flanagan asks readers to imagine a video surfacing online in which their chief executive announces major redundancies, or says something wholly at odds with the company’s values.

The instinct may be to investigate and collect evidence before making any public statement, but she warns: “That instinct can be costly.” When AI is involved, the opening hour carries disproportionate weight, because employees, customers, investors and the media make up their minds long before a formal investigation is finished.

“Silence creates a vacuum, and misinformation fills it,” she writes.

Her priority is to set up a trusted source of facts, communicating quickly through verified channels and ideally through recognised leaders, since people tend to trust individuals more than corporate statements. A plain, factual message from a senior figure can frequently contain a situation better than pages of cautious legal wording, she says.

“Speed and credibility are no longer competing priorities. You need both.”

Synthetic text

While much of the debate about AI misinformation centres on video, Flanagan suggests synthetic text could prove harder to tackle. A fake news article shared as a screenshot takes no technical skill to make believable, as it mirrors how most people now consume news, via social feeds, private messaging groups and algorithmic recommendations.

False claims about a regulator’s investigation, a data breach or misconduct by an executive can spread fast among people who may never see a correction.

Traditional rebuttals often fall short, she argues, and leaders should do three things at once. First, be specific, as vague talk of misinformation seldom works and audiences want facts they can check. Second, bring trusted stakeholders in early so that media contacts, analysts and key partners hear from the organisation directly rather than via a social feed. Third, set escalation routes in advance, because organisations with existing links to platforms’ trust and safety teams can often act much faster once harmful content is circulating.

Preparation ties these together. “Relationships built during calm periods become operational assets during a crisis,” she writes.

Every function involved

Flanagan sees a weakness in crisis plans that still make communications the main owner of the response, a model she says belongs to a different era.

Incidents involving AI touch on reputation, cybersecurity, legal risk and technology as well as employee engagement, and no one function can deal with them alone. In her model, communications owns the narrative and trust, legal limits unnecessary exposure, and cybersecurity determines whether material is real, altered or invented outright. HR handles staff uncertainty and morale, while the executive team provides visibility and authority.

Coordination is the difficulty. Many organisations have never rehearsed an AI-enabled crisis across those teams, she says, yet they may be expected to take high-stakes decisions minutes after an incident appears. How they first respond often decides whether the situation stays contained or becomes a wider reputational problem.

The most effective organisations, she argues, are shifting from reactive planning to resilience. They run simulations involving deepfakes and synthetic media, approve response frameworks ahead of time, and line up external validators, fact-checkers and industry bodies that can lend credibility when trust is being tested.

They also accept that AI risk extends beyond the communications team, with every employee now having a part to play in spotting suspicious content, checking information and protecting trust, which Flanagan calls “a significant shift in mindset”.

The basics are unchanged, she says, and leaders still need to be “fast, factual and human”. What is different is the environment, with AI spreading misinformation faster than traditional decision-making can keep up. The organisations that adapt, she argues, will treat crisis readiness “not as a communications exercise, but as a business capability”.

Written by

Digital Gossip

News, people moves, campaigns, deals and opinion from across the UK’s agencies, marketing, media and technology businesses, written and edited by the Digital Gossip newsroom.

Send us a tip

Read next