AI - Friend or Foe?

Artificial Intelligence (AI) is with us now and furthermore, it is unlikely to go away.  However, belated regulation may be introduced in an attempt to off-set the potential risk that AI may present.  The discovery that a rogue Open AI agent infiltrated the Australian government website and accessed data from Australia’s universal healthcare scheme, Medicare, has caused considerable consternation regarding safety of the personal and sensitive data that could be accessed and used for a range of inappropriate actions, both in commerce and importantly, the damage that may be done in the world of politics where the global risks could be catastrophic.  Forbes reports that one in four breeches are AI enabled.  As well as IBM warning that threat detection is paramount for all organisations.  Warnings are being issued to commercial entities on a daily basis. 

Also, businesses must use the utmost vigilance when using AI in their commercial dealings as the misinformation that AI can produce, known as hallucinations, is extremely convincing with the potential to damage the reputation of the business.  All AI information must be viewed with extreme caution.  There is no doubt that AI can have a beneficial role which should be maximised globally, for example:

• Early identifying treatable diseases at an early stage
• Advancing new life-saving drugs and vaccines
• Tracking pollution to assist commercial companies to lower carbon ommissions
•  Predicting climate changes such as droughts to enable adequate preparation
• Undertaking routine tasks to enable the focus on advancement and development
• Assisting the disabled with tasks and such things as enabling text to be transformed into sign language.

The UN National Security Council, however, shows concern over governance and safety frameworks and there are concerns that the slow human evolution will be overtaken by the rapid advancement of digital superintelligence.  The risks may be mitigated by universal governance that is accepted globally which does not favour any country over and above any other.  Dealing with the potential risks that AI poses may pave the way for greater universal harmony amongst nations.  Governments and businesses must find a way to harness the benefits and mitigate the risks of Al.

Businesses would be wise to adopt the following measures:

• Introduce an AI policy outlining what AI can and can’t be used for.
• Identify the authorised suppliers.
• Clarify the type of information that can be dealt with in AI 
• Verify the authenticity of all information to identify misinformation.
• Establish IP and copyright rules for AI generated material.
• Ensure that human oversight remains in place to validate the accuracy of information AI suggests.
• Remain fully updated on the latest research on a topic, including being aware of AI hallucinations and other forms of misinformation.

However, if a business published AI created material and subsequently discovers that the information was incorrect there obviously can be consequences.  The fact that the error was generated by AI does not as a rule transfer responsibility to the AI provider.  Businesses remain responsible for their publications regardless of the source of the information. 

The liabilities that the business can be responsible for in these circumstances are:

• Misrepresentation, especially if their clients relied on the information.
• Breech of contract, f the information relied could be deemed to be part of contractual commitments.
• Negligence if a third party suffered a loss due to relying on the information.
• False statements that are repeated about an individual could result in defamation.
• Information can sometimes be reproduced that creates a copyright risk or infringement of intellectual property.
•  Professional businesses reproducing inaccurate material may face professional negligence.

The consequential fines resulting from some breeches of consumer law can be substantial.  Also, there will almost certainly be regulatory consequences depending on the industry or professional sector concerned.  The insurance sector appears to be reluctant to embrace the risk AI poses. Most insurers are only prepared to engage in certain aspects that may arise.  The wording of the policy is key, such as the ability of the business to adequately check the AI generated material.

An event in London in October aimed at senior leaders in the AI sector aims to present case studies, workshops and boardroom sessions under Chatham House rules to offer expert guidance on new innovations, growth, infrastructure and more.

The lawyers at Giambrone and Partners can assist in undertaking research into deep fakes and AI hallucinations.

For assistance please email client.services@giambronelaw.com or call 020 7183 9482