The UK’s data protection watchdog has launched a consultation series on how data protection law should apply to the development and use of generative AI apps.

The Information Commissioner’s Office (ICO) said the first consultation would look at whether it was lawful to train generative AI models – such as ChatGPT – on personal data scraped from the web.

The ICO said it was seeking views from across the tech and AI industry, as well as from legal advisers, consultants and civil society groups, with the aim of helping the regulator provide the sector with clarity on how to abide by data protection law when using the technology.

Generative AI technology can be used to create new content – including words, pictures, video, computer code and music – from scratch, but based on large amounts of data the program has been trained on, much of which is often scraped from internet and fed into the system.

Over the last 12 months and since the public launch of OpenAI’s ChatGPT chatbot, generative AI has become one of the most on-trend pieces of technology in the sector, with Google, Microsoft and others quickly moving to release their own chatbots and build generative AI tools into their platforms.

Critics have raised concerns about the number of areas around generative AI, including its ability to create misinformation if it has been trained on inaccurate data, as well as the potential privacy concerns around personal data being used in such training.

The ICO said its first consultation, which will be open to hearing views from stakeholders until March 1, would examine when and if it is lawful to train generative AI models on personal data.

Stephen Almond, executive director for regulatory risk at the ICO, said: “The impact of generative AI can be transformative for society if it’s developed and deployed responsibly.

“This call for views will help the ICO provide industry with certainty regarding its obligations and safeguard people’s information rights and freedoms.”

The ICO said future consultations, which will be launched throughout the first half of the year, would also look at areas such as the accuracy of generative AI results.