AI has opened up a world of opportunities but it also comes with risks - Ben Verinder
Artificial intelligence (AI) is being adopted by individuals and organisations around the world and the pace of uptake continues to accelerate. But with considerable benefit comes significant risk.
In the UK, according to recent research by accountants EY, two thirds of us use AI platforms like ChatGPT or Co-Pilot regularly, with AI helping just under half of us at work, where AI models promise significant increases in productivity by automating or supporting repetitive tasks, freeing us up to focus on more proactive or creative activity.
As well as helping us work faster, AI offers us new capabilities.
In my industry – public relations – these include much more sophisticated measurement of our work, hyper-personalised communication and global, multilingual campaigns at a fraction of their previous cost.
AI’s societal benefits are considerable. It is improving the accuracy of cancer diagnoses, helping councils to predict potholes and supermarkets to minimise food waste.
AI is being deployed for a dizzying range of tasks – from conservation management to toy production, film restoration to missile defence, disease prevention to astrophysics.
But most people using AI in the UK have not been trained on the new technology and most organisations do not have a policy to guide staff use of AI. This presents huge risks to the organisations that employ us.
AI use can quickly reduce the quality of communication between us, and between an organisation and its customers, partners and other important groups.
AI models like ChatGPT are often trained to generate content quickly and in large volumes. These models can struggle to understand the nuances of human language and culture. If they are fed vague, imprecise instructions this can lead to ‘slop’ – low-quality, generic, repetitive content, often characterised by a lack of originality, coherence and a tonal monotony. Slop is a consequence of how models are programmed and trained but, predominantly, how we use them.
The use of AI to generate marketing or communication content raises copyright risks. In the UK, material produced by artificial intelligence, without sufficient human intervention, is not afforded intellectual property rights. Which means that organisations using AI to produce advertising or promotional material are at risk of losing their ownership of that work.
We might also be deploying AI in ways that are unacceptable to the people that matter most to us. A recent study by the Ada Lovelace and Alan Turing Institutes revealed that the British public’s views on what is and is not an acceptable use of AI differs greatly depending on the situation. We are, for instance, most concerned about AI use in advanced robotics, advertising and recruitment and less so about its adoption in education or policing.
Staff who have not been trained to provide clear instructions or ‘prompts’ to AI tools risk generating ‘hallucinations’ - false, misleading, or nonsensical information that the AI model presents as factual.
If organisations take hallucinated content at face value they face considerable risks, including failure to comply with regulations or the law, misinformed decision-making, disruption to operations, security breaches or service failures.
Organisations are also going to have to reconcile the environmental costs of AI – in particular the energy and water used by data centres – with their own environmental commitments. For some, this is likely to be impossible.
Ben Verinder is a specialist in the reputational risks of artificial intelligence and co-editor of ‘AI for PR’, due out in May 2026 from Kogan Page.