Digital safety charity criticises EU over AI Act's 'get-out clauses'

Alex Cooney, CEO and co-founder of CyberSafeKids.ie: 'You could just see the court cases coming down the track.'

European lawmakers have been criticised for the inclusion of 'get-out' clauses in its artificial intelligence (AI) legislation, which it is claimed will give rise to court challenges from the tech's creators.

The chief executive of digital safety charity CyberSafeKids, Alex Cooney, said it "frustrates me enormously" to see AI regulations being watered down.

EU countries and European Parliament lawmakers in May agreed to new rules governing AI, including delaying ​their implementation, in a move which critics said showed Europe caving to big ‌tech.

At the launch in Dublin today of the CyberSafeKids report regarding the digital lives of Irish children in the past year, Ms Cooney said the updated legislation is akin to "get-out clauses".

"Just recently there are these amendments to the AI Act introduced, and they're now requiring that you make it clear in an AI chatbot that you're talking to an AI," she said.

"But actually there is a get-out clause as well, because it's saying 'to a reasonable person', if it's obvious already then they don't need to say it. Well, who's determining what that is?

"You could just see the court cases coming down the track," Ms Cooney said, adding that "we need much more specific measures".

She added that "we need to actually define what 'safe by design' means in the European context" to ensure tech products can be mandated to meet the requirement.

The CyberSafeKids report found that the use of generative AI by children has "exploded" over the past year, with four times the number of 12- to 15-year-olds now using the technology for schoolwork compared with last year, and double the number of eight- to 12-year-olds.

The report also found that 62% of students in DEIS primary schools aged from eight-12 have a smartphone — more than twice the numbers reported in non-DEIS schools.

Meanwhile, access to social media platforms such as YouTube (63%), Snapchat (21%), and TikTok (20%) remains high for children in the same age brackets.

Olwyn Beresford, the lead researcher at CyberSafeKids, said the suggestions around the safe use of AI, such as parental controls, have already been proven to not be a solution.

"Have we not done this already? Have we not been here with social media?" she asked.

"Did we not just let them use it, wait for harm to be proven, and then start floundering and say 'we'll slap on some parental controls, that's going to solve the problem'," said Ms Beresford.

"We need to come up with a solution that will cause change immediately because we can't wait. They're already using all these products." 

Ms Beresford said that, in terms of chatbots usage, "with younger kids they're just chatting, playing". "They're just enjoying it for advice, for company, for entertainment, and that's quite worrying when you consider the answers they're getting back and... the lack of safeguards."

Ms Cooney said the rise of AI chatbots requires "a public health response" and that it is "not easy" for all parents to just deny their children access to technology or social media.

"We need to look at why that is the case so that we can support them. And we can't put all this (responsibility) on parents," she said. "We need to enforce the regulations, but also ensure that the regulation is putting accountability on these companies." 

Ms Cooney said that fining tech giants would not act as a deterrent given their wealth, describing the $18bn penalty levied on Meta last week in the US as "chickenfeed" compared to its resources.