Warnings over AI risks to children 

'Generative AI is present in systems such as chatbots, interactive games and is even being offered more worryingly as a friend on social media.'

Children can place “too much trust” in artificial intelligence (AI) which can amplify the harmful content they see online, an Oireachtas committee is to be told.

Set to address the Children’s Committee on Tuesday, the media regulator Coimisiún na Meán will warn that while AI has become an increasing feature of children’s lives, it “was not designed for children nor does it use a safety by design approach”.

“Generative AI is present in systems such as chatbots, interactive games and is even being offered more worryingly as a friend on social media,” the Coimisiún’s online safety commissioner Niamh Hodnett will tell TDs and Senators.

“While AI can help children learn and play, it poses risks — children may place too much trust in AI systems, AI may provide unsafe or false information, children can come across content that is age-inappropriate.

“There are growing concerns in relation to AI-generated content, particularly the manipulation of imagery through deepfakes, and AI-generated child sex abuse material.” 

Regulation

Last week, the Oireachtas Children’s Committee heard that Ireland is “not doing enough” to protect children from the dangers of AI despite the potential to be a leader in this space. Members were told that children were being “treated like guinea pigs in the digital world”.

It heard that in the Data Protection Commissioner and Coimisiún na Meán, the country had the regulators with the teeth necessary to clamp down on bad practices from tech giants, but enforcement must be stepped up.

Coimisiún na Meán is responsible for regulating services which have its EU headquarters in Ireland, which includes Meta and X, formerly known as Twitter. 

Last week, Ireland’s media regulator welcomed the EU Digital Services Act coming into force. Together with other pieces of legislation, it grants Coimisiún na Meán powers to hold digital platforms accountable for how they protect people, and especially children, online.

Ms Hodnett will tell the committee that measures around AI were included in the regulator’s draft Online Safety Code aimed at protecting children from harmful content.

“AI-driven recommender systems can present risks including the amplification of harmful content, the recommendation of age-inappropriate content, disinformation, the facilitation of inappropriate relationships between adults and children, and excessive amounts of time online,” she will say.

“One of the supplementary measures we have proposed is for platforms to take safety measures to reduce harm caused and to conduct a safety impact assessment.

“If AI is to work for children, children need to be front and central in its design.” 

What's your view on this issue?

You can tell us here

Ms Hodnett will also say that while AI can be useful for content moderation, it can also make “inaccurate or biased decisions”.

However, AI could be used to aid age verification techniques which the commission says must be robust to protect children from content that is inappropriate for their age.

Ms Hodnett will conclude: "I want to assure you of our commitment to use our functions to serve children well."