AI investment lacks basic safety checks, warns BSI
Many businesses are investing heavily in AI without sufficient safeguards for its implementation, a new study by the British Standards Institution (BSI) has found.
While executives are touting AI’s potential to drive productivity, the report suggests that most firms are operating with little more than blind faith in a technology they do not yet fully understand.
The research also found that 62 per cent of business leaders plan to increase AI investments over the next twelve months, citing efficiency and cost savings.
But fewer than one in four firms (24 per cent) have an AI governance programme in place. Among larger firms, that figure only rises modestly, to a third.
BSI boss Susan Taylor Martin told CityAM: “The business community is steadily building up its understanding of the enormous potential of AI, but the governance gap is concerning”.
“AI will not be a panacea for sluggish growth or low productivity without strategic oversight and clear guardrails. Overconfidence, coupled with fragmented and inconsistent governance, risks leaving many organisations vulnerable to avoidable failures and reputational damage”, she added.
Governance lags as risks rise
The report comes amid ongoing conflict around how governments or regulations should balance prioritising innovation and maintaining accountability.
A few weeks ago, the UK government launched plans for its ‘AI growth lab‘, a regulatory sandbox designed to let organisations experiment with AI under tighter controlled conditions.
While the initiative was hailed by industry leaders, concerns arose that safe experimentation should be grounded in strong ethical oversight.
The BSI study found that just 28 per cent of executives know which data sources their businesses use to train or deploy AI tools, a 35 per cent decline from earlier this year.
Meanwhile, just two-fifths said their business had clear protocols in place for handling confidential data used in AI training.
Across all sectors, risk management appears patchy, with nearly a third of executives saying AI has been a source of risk within their businesses.
However, only 30 per cent reported having a formal risk assessment process, with even fewer, 22 per cent, saying their firms restrict unauthorised employee use of AI tools.
Human skills take a back seat
Despite the heavy-handed investment, the BSI’s analysis found that the term ‘automation’ appeared seven times more often than ‘training’ or ‘education’ did in executives’ annual reports, reflecting a wider neglect of workforce preparedness.
While over half of executives say they are confident their staff have the skills needed to use AI effectively, only a third have developed a training programme accordingly.
Taylor Martin warned that “businesses may be underestimating the importance of human oversight and capability alongside tech advancement”.
The study follows similar findings from Cisco, which last month found that only 16 per cent of UK businesses are ready to scale AI safely.
“AI infrastructure debt is the build-up of small shortcuts that quietly add up”, Sarah Walker, Cisco’s UK chief executive, told CityAM. “Over time, they slow progress, raise costs and limit the impact of AI investments”.