Business founders don’t trust AI, yet half are leaking sensitive data anyway, Startups report reveals

AI is prevalent in UK startups right now, and so are anxieties around its use. Half of business founders have shared sensitive information with public AI tools in the past month, despite the vast majority expressing fears around the privacy of these models, our new survey, in collaboration with Sage, finds.

Startups looking to sidestep the risks of public AI with custom stacks are also facing roadblocks, with 62% wasting AI budgets on fixing broken connections between isolated tools. This is leaving business founders with a dilemma: accept the risk, or absorb the costs.

To help you navigate this quandary, we’ve proudly partnered with Sage to release the definitive 2026 report and playbook for founders: ‘The Startup AI Paradox’.

Building on the survey results of over 400 business founders, owners, and C-suite executives from UK startups, this report unpacks issues being felt by current business owners using AI, and explores potential solutions. This is what we’ve revealed.

 90% of SMEs using AI worry that private data is being used to train global models

As startups increasingly lean on public AI tools for quick productivity wins, one in two business founders admit to leaking sensitive company information with chatbots like ChatGPT and Gemini in the last 30 days, and for over a quarter of them, it’s becoming a weekly habit. 

This isn’t down to naivety, though. 90% of AI users are concerned that the private information they enter into tools could end up training someone else’s model, and over a quarter (27%) say they have little-to-no trust in AI handling core business processes. 

The risks are tangible. Once sensitive company information is entered into a public model, it can’t be recalled, deleted, or contained. This can result in client data resurfacing in another user’s output, as well as potential GDPR penalties. 

This gap is being driven in part by competitive pressure. As the business landscape rushes to fold AI into their daily workflows, standing still can feel like an ever bigger risk. Our survey also found that 83% of business leaders feel overwhelmed by the number of options available, resulting in many of them defaulting to whatever’s available. 

Whatever’s informing their decision, the findings lay bare the fact that founders know the risks of using public AI, and they continue using it anyway.

Businesses face a catch-22 when it comes to choosing how to use AI

For founders drawn to the speed and accessibility of public AI, using it safely isn’t impossible – it’s just burdensome. Public tools offer no data security guarantees, requiring founders to anonymise sensitive figures, scrub internal documents, and remove confidential details before a single prompt is sent.

Our survey results highlight how tedious business founders find this process, with over half (56%) of respondents spending a moderate amount of time on manually preparing data, and almost a quarter (23%) finding it completely self-defeating.  

To avoid these issues, some businesses build their own AI tech stack, where each model can handle different information streams, from customer data to finances, giving founders more control over where sensitive data goes.

However, while a step in the right direction, company-made AI infrastructure can increase costs as quickly as they improve efficiency, with 62% of businesses with dedicated AI budgets reporting wasting significant time trying to fix broken connections between their tools. 

Ultimately, whether businesses absorb the risks of public AI tools or invest in fragmented custom stacks, trade-offs remain. Neither path solves the problem, but our findings point to another alternative.

 Get the full picture with our expert-led AI playbook

Whichever stage your company is at in its AI journey, there is help at hand. Our free report, “The Startup AI Paradox”, acts as a comprehensive AI playbook for founders navigating the trade-off between speed and security. 

It features insights from Startups 100 winners, a starter checklist for founders looking to create an AI policy, and legal guidelines explained in clear English, all designed to remove the guesswork around adopting and thriving with AI. 

We also look at the solution that could resolve this paradox for good. Download ‘The Startup AI Paradox’ for free and get the complete findings from our survey in collaboration with Sage.