PR is rooted in networking, storytelling, and empathy. All three need active listening, adjusting based on audience feedback, and value creation, providing meaningful insights or telling compelling stories that build trust and loyalty. And AI is now driving this role like never before.
Beyond automating manual tasks, AI, first with LLMs and now Agentic AI, has helped unshackle PR professionals to get back to what they love doing most: landing big media hits, pitching stories creatively, impacting change, and turning seemingly disconnected developments into major story opportunities for clients, and more. Not surprisingly, AI now sits at the heart of PR workflows. It drives ideation, narrative development, hyper-personalised pitching, real-time crisis monitoring, and performance tracking, transforming brand management like never before.
Yet, using AI without guardrails and human oversight carries significant risks to the client, the media, and the agency, introducing reputational hazards that can erode brand trust, lead to customer churn and financial liability. Before dwelling into the risks and how to address them, the following section will take a look at how AI works and why oversight is necessary in PR workflows.
Understanding use of AI in PR and Risk
AI is omnipresent in PR already. Research, brainstorming, translations, editing, proof reading, transcription, summarising meeting notes, sentiment analysis, social media monitoring, generating text, images, moodboards and videos, and even replacing the PR professional completely, AI is layered into PR workflows now. This power has brought speed and sheen transforming agencies irrespective of size into proactive, data-led, creative workhorses.
At their core, LLMs, such as Chat GPT or Claude that PR teams use extensively, are built on deep neural networks and trained on massive data sets of text to perform tasks like answering questions, or translation languages. LLMs are defined by their ‘parameters’, millions of them, which gives them capacity to grasp nuanced language, idioms, and vast stores of stored general knowledge. LLMs predict what should come next based on probabilities and not actual facts. Which means they can confidently invent or imagine incorrect information. And because they learn from vast amounts of human generated data on the Internet, they do carry bias embedded in the underlying data they reference unintentionally.
Therefore content generated by AI tools, whether text, image, simulations, or videos, can mislead in many ways.
PR is really the art of perception management. By shaping the narrative, influencing how the public understands and engages with a brand, it is possible to drive targeted response in a way that suits the brand. It is likely that unchecked use of AI in PR workflows or to influence AI searches can risk brands going off message, parroting competitors unwittingly, and lead to reputational loss. Furthermore, it threatens to pollute the underlying information ecosystem, and create legal challenges or unknown algorithmic liabilities.
Current State of Play
Journalists already use AI detection tools to validate human input, demanding direct spokesperson interactions to validate source authenticity. A few even disclose the use of AI tools in their work when so permitted by their organisation. Media organisations, on their part, have implemented strict guidelines, regarding use of AI in day to day work, feeding confidential source information to AI tools, image creation and autonomous publishing, to prevent inadvertent misinformation and to protect brand credibility.
On their part, brands and PR agencies have begun implementing guidelines to ensure AI is used for drafting and ideation and then deploying human oversight, to ensure brand tone, sensitivity and to stay within agreed brand messaging pillars.
Guardrails for use of AI in PR
- Use Enterprise Versions of AI Tools to prevent data retention for tool training purposes
- Redact/mask client names and other sensitive details before submitting inputs to AI tools
- Implement a layered approval process to ensure the final output is human verified with multiple functionaries, with graded roles for fact checking and ensuring legal protections, reviewing the output before it is released to the public.
- Continuously educate users on the team regarding the AI tool limitations and how to identify bias and tool hallucinations.
- Implement legal clauses in Master Service Agreements (MSAs) regarding who owns the AI generated content and who bears the liability if an AI related error occurs. This is to protect both the agency and the brand against unintended consequences.
- Decide when you must tell the client or the public about the use of AI
- Ask team to investigate sources till they find a verifiable entity (live interview, corporate document)
AI is laying out an exciting path for the PR industry. Beyond basic generative capabilities, AI will usher in predictive intelligence, cognitive narrative mapping, and optimisation of AI search engines in the quest for a brand’s public perception management. Rather than replacing the man in the middle, AI will empower PR professionals and agencies from tactical operatives to the gatekeepers of a brand’s image and reputation.
Written by: Dayanand Muthukrishnan, Founder and CEO at EXL Public Relations.