AI Joyriding - A Wake-Up Call to Set Guardrails Before Hitting Scale

We all know that the AI revolution has well and truly arrived. From ChatGPT writing emails in seconds to Canva’s Magic Studio creating social content with a single click, AI is no longer merely coming, it’s already embedded in the way we work. And while the opportunities are vast, so too are the risks, especially when we don’t slow down long enough seriously consider whether we’re applying AI correctly to our existing working ways.

So just what is this potentially maverick - and possibly irresponsible sounding - approach to adopting AI, I hear you ask. AI Joyriding is a term I’ve coined that refers to when employees, freelancers or even leadership teams start using AI tools freely, and without guidance, governance or oversight. They mean well, by saving time, reducing admin and improving output, but they’re essentially driving without a licence, using unregulated AI methods which may result in unclear results, policies, training and strategic oversight.

For me, it’s the intern who pastes client content into ChatGPT verbatim without checking for tone, accuracy, plagiarism or data sensitivity. It’s the recruiter who experiments with AI shortlisting tools without understanding bias or data sharing risks. It’s the manager who pastes a confidential client brief into an AI chatbot, forgetting that these tools do retain and learn from inputs. In fact, large language models use machine learning to feast on the information we feed it, which is why we need to be careful of its capabilities.

While enthusiasm towards embracing AI is welcome, the risk lies in getting results speedily but without any sort of strategy, in turn rendering them useless. It’s also important to acknowledge that joining AI joyriding can be all too easy, particularly in the field of marketing, which is constantly evolving and gathering pace. Unsurprisingly, this makes it almost impossible for businesses to successfully keep up with the latest AI developments and their implications.

If you're a business leader or marketer it’s time to take notice. This isn’t about shaming people. AI Joyriding is happening in every organisation, right now. The question is: Do you have any idea to what extent or where or even how?

So, let’s get stuck in. I’m here to guide you through the concept of AI Joyriding, by exploring its risks within marketing-led businesses and also to offer practical steps to establish guardrails - all before the use of AI becomes chaotic or damaging to a brand’s reputation, its people or performance.

Why It Matters – Especially in Marketing

Marketing, by its very nature, touches every part of a business; from brand voice, data privacy, client experience, employer brand, trust and integrity. That’s what makes it one of the most vulnerable sectors to unregulated AI use. This is because AI threatens consistency if misused. For example, using AI-generated content without human nuance dilutes trust and tone in a world where clients expect authenticity, not automation for automation’s sake. The fact is that AI misuse can erode credibility before a business even realises it’s already happened and that it’s too late to go back. After all, who could forget the demise of Clip Art in the 2000s? This occurred when the advent of search engines and specialised websites offering high-quality stock images and illustrations led users to these resources instead of relying on the limited and unsophisticated clip art collections within software like Microsoft Office.

Let’s look at two prime examples:

• Doing it Right: Monzo – The digital bank has been open about using AI in customer service and fraud prevention. But crucially, they also invest in transparency. Their tone of voice guidelines are public, and they’ve trained their AI to reflect human nuance in that it’s polite, conversational and, most importantly, not robotic. AI supports their brand but doesn’t define it.  In the main, these AI customer service agents are still a bit too ‘un-human’ for my liking, but they're getting better and at a rapid pace.
• Doing it Wrong: CNET – The tech news site quietly rolled out AI-written articles, only for this to be uncovered by readers. Worse still, many pieces had serious factual errors. The backlash damaged trust, not just in the publication, but in the value of AI-generated content more broadly.

For marketers, the lesson is clear: AI isn’t a shortcut to strategy. Yes, it’s a powerful tool, but without human oversight, it can do more harm than good.

Where AI Joyriding Originates

It’s all too easy to see how it happens:

• Tight budgets equate to a lack of well-thought-out, formal AI adoption frameworks, policies and processes.
• AI ‘FOMO’ results in employees taking matters - often well-intentioned but sometimes misguidedly - into their own hands.
• Overstretched teams and resources means there’s little time to vet tools properly and offer the necessary training to implement and operate them at full capacity.

Suddenly, you’ve got dozens of unchecked AI experiments happening across the business, with no shared learning, tracking or awareness of the risks involved.

Defining the AI Joyrider

AI joyriders are to be commended on how tech curious they are with their appetite to save time - a positive slant to the joyriding phenomenon.

However - and through no fault of their own - they often act without fully understanding the ramifications of adopting AI too hastily. After all, the privacy laws, bias risks and data governance procedures which shape the world of AI are so complex that few people can even begin to navigate them under their own steam. Without structure, these so-called joyriders run the risk of exposing their business to reputational, legal and ethical risks.

Which Lead Us to the Realities of AI Overwhelm

AI is everywhere now, which represents a world of exciting opportunities. But for many SMEs and marketers, it’s also overwhelming and confusing, as they feel the pace of change and subsequent pressure to act upon it to remain competitive. They might feel the need to integrate AI into existing systems and workflows which can be complex, requiring specialised knowledge and resources. If tools are adopted without the right training and understanding, this can result in chaos. 

The best way to deal with AI overwhelm involves resisting the urge to overcome the pressure and instead taking steps to adopt AI in methodical ways, without rushing the process and, instead, accepting it takes time. Businesses should allow themselves to adjust to the new reality of AI, by openly communicating about it across all functions, areas and departments to avoid confusion and to build a clear picture when it comes to defining those all-important next steps.

The Risks: From Data Breaches to Brand Damage

Here are just a few of the ways AI Joyriding can go wrong:

• Data security breaches: Data security breaches can manifest in various forms, including cyberattacks like ransomware, physical theft of devices or records, and vulnerabilities in software or APIs, leading to unauthorised access, data loss, or disclosure. Uploading personal or client data into third-party tools can also violate GDPR.
• Inconsistent brand voice: Inconsistent brand voice manifests when a company's communication style varies across different channels, creating a confusing and unaligned brand image. This can include mixed messaging, by using a formal tone on social media but a casual tone in email, or using different visuals, like logos or colour schemes. Aside from that, AI-generated content can sound off-brand or impersonal without human editing.
• Legal liability: AI legal liability arises when AI systems cause harm, and can stem from negligence, breach of duty or violating laws like data privacy and intellectual property. Misuse of AI tools without understanding their terms or limitations can lead to compliance failures, negligence and a lack of duty of care.
• Bias amplification: AI bias amplification occurs when AI systems, trained on biased data or designed with inherent biases, reinforce and magnify existing societal inequalities, leading to discriminatory outcomes. AI that’s left untrained or unchecked can reinforce stereotypes, especially in the realm of recruitment or HR.

  ---

Practical Guardrails to Slow Down the Joyride

You don’t need to be an AI expert to start managing the risk. You just need to implement the right guardrails. Here’s how:

1. Create an AI Use Policy

Lay out what’s acceptable and what’s not. Clarify which tools are approved, how data should be treated, what’s off-limits. This might start with extending your IT policy and employing AI experts, such as Chief AI Officers, AI Engineers and Data Scientists.

• Tool Tip: OpenAI’s usage policies are a good reference point when setting boundaries.

2. Map Out Your AI Activity

Take the time to complete a comprehensive audit. Who’s using which AI tools are used in your team? For what purpose are they being applied to tasks? Are they storing sensitive data? Is content being fact-checked?

• Tool Tip: Use Notion or Airtable to create a shared AI tool tracker across teams.

3. Establish a Review Process

No AI-generated content or campaign asset should go live without human eyes on it. Build in a layer of editorial or compliance review to determine automated data extraction, pattern recognition, plagiarism checks and contract and document reviews.

• Tool Tip: Airtable helps by keeping track of projects and tasks across departments. While managing budgets and automating reporting.

4. Appoint AI Champions

Identify people across departments who aren’t just tech experts, they can be your go-to contacts for ethical, effective AI use, as well as brand stewards. Train them to spot risk and share best practices.

• Tool Tip: TrustLayer platform’s AI-driven risk management tools can help AI champion their job well as they are designed to automate document collection, tracking and verification as well as cross-check documents in real-time, flagging missing information or gaps against the enterprise’s requirements.

5. Invest in Training - Not Just Tools

Before spending thousands of pounds on shiny new AI automation tools, spend a fraction of that amount on team training. A solid foundation of AI literacy is your first line of defence. Earlier this year, Prime Minister Keir Starmer even unveiled a £7 million fund to boost small businesses' productivity and efficiency through AI.

• Tool Tip: Platforms like AI for Good or DeepLearning.AI offer beginner-friendly training resources.

  ---

The AI Sandwich Model (Yes, Really)

Here’s a simple way that I like to explain the layers of safe AI adoption:

• Top Layer – AI innovation. New tools, capabilities, ideas.
• Middle Layer – Governance. Policies, training, guardrails to gain powerful but safe knowledge.
• Bottom Layer – Joyriders. The team, curious and eager, often acting before thinking.

The middle layer is what makes the sandwich work. Without it, everything falls apart.

Closing Thoughts: We Don’t Need to Fear AI. But We Do Need to Manage It.

AI has a huge role to play in any business, but how we use it will shape whether it adds value or becomes a liability moving forwards.

As marketers and business leaders, it’s on us to move beyond excitement and hype and instead towards stability and structure. That means:

• Creating space for curiosity, but
• Backing it with policy, review, and training

We need to empower our joyriders to become responsible drivers, because when we slow down, we can go even further.

AI doesn’t need to be feared, but it absolutely needs to be managed. And it’s not about stifling innovation, it’s about turning enthusiasm into excellence.

Curtailing AI Joyriding should be managed via a ‘risk register’ in which the tools and testing phase is about determining or defining what you feel is acceptable. This might encompass what tasks is AI solving and what the implications are around using AI positively. Without a destination, a roadmap and a safe approach, there will be disaster ahead. Define your risk register and agree as a company what you're comfortable with and have fun! 

In my view, there is a wonderful window of opportunity at this very moment in time to slow down to speed up for successful AI implementation. This will help businesses to harness AI’s full potential without losing what makes their business human, trusted and intentional.