The arrival of AI is often likened to a present day Industrial Revolution: a sudden technological development that changes how we live and do business so dramatically that we can’t imagine what life was like before.
AI undoubtedly brings positives across industries and sectors. From advances in medical care to creating climate solutions, from streamlining inefficient workstreams to offering an opportunity to rethink entire business models, the benefits of AI are being discovered and put to use everyday.
For businesses that invest in innovation, AI tools offer transformational opportunities.
At the same time, though, such an unparalleled moment of change deserves a moment of pause.
As the AI-rush continues, and as murmurs about the AI bubble bursting grow, it is important to think about the potential risks and problems AI might bring for businesses.
These are relevant not just for innovative SMEs and start-ups who use AI mechanisms for disruptive potential, but for larger corporations who, according to recent research by Gartner, are using AI to address bigger picture questions about their businesses’ strategy and underlying philosophy.
AI may well bring with it opportunities to develop structural resilience but it is important to have an awareness of the risks the new technology brings with it.
What are the legal risks of using AI?
LLM tools like ChatGPT or Claude can be useful for drafting letters, pitches and contracts. But businesses should not forget the importance of keeping client and customer data confidential.
According to IBM, in 2025 the average data breach cost globally was $4.44 million. It goes without saying that lack of oversight about confidential data and GDPR use can be costly, especially to SMEs.
So, how do we prevent employees from leaking confidential data?
There is the risk that AI can prompt decisions or suggestions that inadvertently cause data leaks. In March 2026, for instance, the Guardian reported that a Meta AI agent’s suggestion led to confidential company and user data being exposed for two hours to an internal team.
At a recent Plus X Innovation AI roundtable, Simon Ranyard, MD for North West Europe for Orange Business discussed how important it is to shift company's LLM use to their own closed AI environment. This avoids potential leaks which are more likely when relying on open-source agents like ChatGPT.
Businesses should certainly consider whether the risks of open-source agents outweigh the potential expense of an in-house client.
CIOs and data managers should also consider where AI-related data is being stored. While hyperscalers may offer quick solutions, in a world where geopolitical tensions may define technological and server access, private infrastructure and data storage may be a preferred solution.
Can AI hallucinate or give incorrect advice?
Quality over Quantity: The Importance of a Human Touch
At our Plus X Innovation AI roundtable, Simon Ranyard also spoke about the importance of selecting the right AI tool for the job.
This helps to ensure the quality of AI outputs.
Indeed, another potential issue is that as AI and LLM tools develop, AI agents are now learning from other AI outputs.
This can result in a cannibalistic cycle with AI feeding on its own products to develop. Importantly for businesses, this can lead to poor quality results even if prompts are accurate and specific.
If poor quality AI-outputs are used across workstreams and data sets, businesses may experience not just inefficiency problems but potential reputational damage.
It is key for leaders to keep AI-use accountable. This can be done by asking: what is the purpose of AI for a given task? And how do we ensure there is human oversight?
If these questions are not asked, overdependency on AI-tools may grow and this can again lead to poor quality work.
An overdependency on AI could therefore lead to unintended errors and costly inaccuracies, not to mention a generational depletion in workforce skillsets.
Businesses may benefit from a regular SWOT analysis of their AI use and cost. Given the proliferation of AI tools, too, risk arises from potential outlay.
Though some software companies, such as Adobe, have shifted to an outcome-based pricing model for their AI-tools, SMEs especially may benefit from a robust cost-benefit analysis to cut down on overspending and work-duplication.
It is clear that the success of AI will always require a human touch and, as Simon Raynard said at our roundtable, that this new technology is an accelerator of innovation rather than simply an end goal in its own right.
At Plus X Innovation, we offer a range of corporate innovation opportunities that can help you and your business mitigate the risks of AI in your company. From sprints and hackathons to scaling and concept proofing, no matter the stage of your journey, we provide tailored advice and help.
One thing is certain: AI cannot take away from the importance of meeting and networking with other innovation leaders.
At Plus X Innovation, we offer work spaces and state of the art facilities for product development as well as a range of networking and panel events which unpick the big questions affecting the innovation sector.
Get in touch to organise a tour of one of our Plus X Innovation Hubs in Brighton or Slough.
How can we help?
At Plus X Innovation, we offer a range of corporate innovation opportunities that can help you and your business mitigate the risks of AI in your company. From sprints and hackathons to scaling and concept proofing, no matter the stage of your journey, we provide tailored advice and help.
One thing is certain: AI cannot take away from the importance of meeting and networking with other innovation leaders.
At Plus X Innovation, we offer work spaces and state of the art facilities for product development as well as a range of networking and panel events which unpick the big questions affecting the innovation sector.
Get in touch to organise a tour of one of our Plus X Innovation Hubs in Brighton or Slough.
.jpg)
.jpg)
.png)
.jpg)
.jpg)
.jpg)
-compressed.jpg)
.jpg)
.jpg)
.jpg)
-compressed.jpg)
.png)
-medium.jpg)
.jpg)
.jpg)
.jpeg)
.jpg)
.jpg)
