AI: Business risk, friend or foe?

The integration of AI tools into business tasks is gathering pace, but is this rapid adoption exposing too many organisations to too much risk? The answer may well be ‘yes’. Many companies have not created AI usage policies. This is despite AI tools now commonly being used by employees, whether to summarise a document, draft reports, carry out research, or create social media posts. Whilst adoption is relatively simple, understanding how the tools work is a mystery to many users.

This needs to be quickly addressed. Whilst AI regulation is currently only through EU law, AI usage could breach many existing regulations here in the UK.

How do AI models work?

Within generative AI, algorithms generate or create an output such as code, data, text, or imagery. They do this based on datasets on which they have been trained, which are never up-to-date in real-time. The output is only as good as the user’s input, whether that is a prompt, instruction, or a document or data for bot analysis.

Therein lies a major risk. The AI model can use your material to build its own knowledge and potentially then share it with other users. Your unique insight could easily answer another user’s prompt query. Your business plans could be made public. If input information contains personal customer details, it could enter the public domain. That would be deemed a GDPR breach. But do employees even know to tick the model’s privacy button? Can this be trusted to actually ensure privacy?

AI and intellectual property

The other danger area is intellectual property and copyright. AI models are trained by collating billions of text files, videos, and photos. Much is protected by Intellectual Property rights, copyright law, and trademarks. By simply using information produced as an AI bot output, an employee could infringe the owner’s rights.

High-profile lawsuits are already underway. Disney claims one AI image generator has become “a virtual vending machine, generating endless unauthorised copies of Disney’s and Universal’s copyrighted works”. It points to “a bottomless pit of plagiarism”.

Gaining any copyright infringement compensation from an AI provider is highly unlikely if using a free version of the model. A paid version might offer some recompense, but businesses should review the terms and conditions.

The need for human input

One of the key issues with AI is the over-reliance users place on output. Unless a human verifies material, assesses whether it could infringe copyright, and checks the facts, mayhem can result. AI models also contain bias, due to the way they are trained. This can again cause legal issues, should it, for instance, lead to biased decision-making in recruitment.

Reputational damage is a considerable risk accompanying the use of AI models at work, but there are significant time-saving advantages. Adopting AI correctly should mitigate the risks attached. This starts with thorough staff training, encompassing how the tool works, its risks, and guidance on intellectual property and copyright laws.

Putting AI policies in place

A robust AI policy, explaining permitted and non-permitted uses of AI and what an employee is and is not allowed to input, should back this. It should carry a clear data security statement and detail required verification processes. The employer’s expectations, in terms of honesty and transparency, regarding whether an employee uses AI within a task, should be clarified.

As many employees are already using AI, most employers will have to start the risk reduction process by carrying out a thorough AI audit to assess how and why AI is currently being utilised in the workplace and by whom.

The message is to take a considered approach to AI, rather than rushing to embrace it unprepared. It has an admirable thirst for knowledge acquisition, but if you feed it sensitive or unredacted material, that could come back to haunt you. And do consider the insurance that could protect you if you find yourself in a legal battle. We have a range of solutions that can be your safety net. Get in touch with L Wood Insurance Brokers by calling 01274 515747