Businesses need to beware of 'ghost AI' use by employees: Andrew Brindley

The first is maximising opportunity and productivity. AI literacy training enables SMEs to benefit from technology instead of watching larger firms capture the value and talent. SME business leaders should be leading the charge, mapping out where AI can add value and making sure their teams are equipped with the skills they need to realise that value.

Often, this involves a mindset shift. When a new tool is introduced, it’s human nature to treat it in the same way as an old tool. We see a lot of people treating large language models in the same way as they would a search engine. We need to reframe the way we think about AI.

LLMs are reasoning models and learning how to use these capabilities to the full will help SMEs find newer, better processes that will improve productivity. Unless people are trained to use LLMs effectively, businesses will not capitalise on the benefits.

Research from Anthropic – the team behind the Claude large language model – supports this.

Anthropic has analysed the way customers are using Claude and mapped actual use across different industry sectors against the LLM’s full capabilities. The findings make for interesting reading and highlight huge gaps in AI adoption, particularly across service sector businesses.

In the legal sector, for example, where security and safeguarding are of the utmost importance, Claude has the capability to manage around 90 per cent of tasks, and adoption currently sits at around 20 per cent. The message is clear: armed with the right knowledge, and with the right safeguards in place, there is significant opportunity for businesses to improve systems and boost productivity.

The second AI literacy challenge makes for more sinister reading. Research by Microsoft found that around 71 per cent of employees have used unapproved consumer AI tools at work, and just over half continue to do so weekly.

So, while you may not be adopting AI use as an organisation, many of your employees are using it. This applies to employees of all ages and seniority levels, and the likelihood is they are using an entry level version with limited privacy safeguards.

Known as ‘shadow AI’ or ‘ghost AI,’ this presents a huge security risk. Ghost AI should be on the risk register of every business, along with the associated intellectual property and cyber security risks.

The government has identified a £400bn AI skills gap and introduced programmes to build AI literacy in the UK workforce. AI adoption and literacy belong on every boardroom agenda.

Upskilling your team is not only important for maximising productivity, but also a strategic business investment in risk mitigation.

Andrew Brindley is co-founder of Conker AI, an AI training and consultancy helping SMEs adopt AI with confidence