Why your business shouldn’t blindly embrace AI

As generative AI and machine learning continue to evolve, cybersecurity companies are warning businesses about blindly adopting these technologies.

According to Ramsac, a cybersecurity provider, large language models (LLMs) can be detrimental to companies if used improperly.

LLMs rely on accessible data from the open internet to inform queries and responses for users, making data privacy difficult to protect.

If you’ve ever used ChatGPT, you would have noticed it doesn’t ask for your privacy preferences before use. This means LLM systems like ChatGPT can’t decipher between confidential and readily available information, meaning intellectual property or commercially sensitive details could easily be exposed.

These concerns escalated to such lengths that Italy temporarily banned ChatGPT in April, after OpenAI was accused of unlawfully collecting users’ data.

Although ChatGPT eventually went live again in the Mediterranean country, it hints at the dangers of blindly jumping on the AI bandwagon.

The dangers of blindly using AI

Whilst AI has been used for transformational projects like finding new drugs or creating robot skin that resembles human touch, it has also been used for malicious purposes.

Although the full extent of cybercrime is yet to be determined, generative AI is already being used to launch sophisticated phishing scams. Cyber attackers can use the technology to script and automate communication without spelling errors, making them less suspicious.

According to a report by TrendMicro, the features that make AI and machine learning systems integral to businesses are also the very same features that cybercriminals abuse for ill gain.

Whilst research shows that AI technology has not represented a lower barrier to entry into cybercrime and therefore, more cybercriminals, it is making their methods more sophisticated.

With cybercriminals taking advantage of AI, it is crucial to be prepared to integrate the technology into daily business routines to prevent being compromised.

To prevent walking into a cybertrap, there’s a few things that businesses can do:

• Avoid using public LLMs for business-specific tasks or information, such as reviewing redundancy options
• Use an LLM from a cloud provider or self-hosted as this offers more security and privacy
• Consider the queries and requests before submitting them to LLMs as it’s possible for this information to be hacked and leaked
• Avoid including sensitive information on public LLMs, such as confidential data
• Submit business critical queries on private or self-hosted LLMs only
• Ensure up-to-date cybersecurity monitoring is enabled and active so breaches and threats can be detected

Future-proofing the workforce

Being well-versed in LLM usage should, therefore, be a priority for businesses that want to integrate AI into their operations. However, doing so will require bridging the digital skills gap that currently weakens the UK workforce.

According to research by Salesforce, only one in ten workers have key AI skills. Moreover, of the 11,000 employees that were surveyed, a scant 14% said their role involved related digital skills like encryption and cybersecurity.

If businesses are to protect their information from unwelcome eyes, it’s paramount to train employees to use new technologies and systems that are inevitably being normalised in the office.

This is key for the existing and future hires. According to research by AND Digital, 20% of workers did not apply for a job and 26% did not seek or achieve a promotion because of their lack of digital skills.

Conversely, research has revealed that 79% of small business owners in the UK consider the adoption of new technologies to be critical for future growth. However, 29% said a shortage of skilled workers poses a high or very high risk for their business.

Therefore, businesses that commit to upskilling employees and equipping them with the in-demand arsenal of digital skills are more likely to benefit from business growth unhindered by data leaks and cybercrime