---

FOI act could cover use of AI tools – MHC

Lawyers at Mason Hayes & Curran (MHC) say that the use of generative AI tools such as ChatGPT by public officials could come within the scope of Irish legislation on freedom of information (FOI). 

They were commenting on the release of records earlier this year showing that Britain’s technology secretary Peter Kyle had used ChatGPT to support his ministerial work. 

The interactions were disclosed in response to a request from New Scientist magazine under Britain’s FOI legislation. 

The request was initially refused on the basis that the information would include conversations made in a personal and official capacity. It was granted, however, when the magazine clarified that it was seeking only those made in an official capacity. 

Records can be ‘held’ anywhere 

In a note on the firm’s website, the MHC lawyers write that there is no specific guidance on whether Irish FOI legislation can be used to access public officials’ ChatGPT records. 

“However, these records could be in scope if principles from the current FOI legislation, and FOI guidance on non-official devices, are applied,” they add. 

The lawyers cite Government FOI guidance that emphasises that records can be ‘held’ anywhere – including in non-official systems, email accounts, and devices. 

These records are subject to the Freedom of Information Act 2014 where:

• They relate to the official functions and/or business activities of a public body, and
• The public body has a legal right to procure the records, regardless of whether they are held in official or non-official systems. 

‘Carefully manage use’ 

“Applying these principles, if ChatGPT and other AI tools are used to generate or inform work-related content, these interactions could be accessible in Ireland under the FOI act,” the MHC note states.  

“As with all FOI requests, it will depend on an analysis of the record’s contents and the relevant facts and circumstances,” the lawyers add. 

MHC points out that recent Government guidance on AI advises against the use of generative AI in the public sector, unless based on an approved business case, and against the use of ‘free-of-charge’ generative AI tools, such as ChatGPT’s free version. 

The firm’s lawyers urge public-sector bodies to “carefully manage” the use of AI within their existing information governance and other relevant governance frameworks, both generally and for FOI purposes.