AN UPDATED DRAFT Online Safety Code has been criticised for not addressing “toxic” algorithms.

Coimisiún na Meán, Ireland’s online and broadcast media regulator, today published an updated draft following public consultation.

The final code will set binding rules that will apply to video-sharing platforms who have their EU Headquarters in Ireland.

Coimisiún na Meán has submitted the Code to the European Commission and once that process is complete, it will be applied later this year.

The final Code will be part of Coimisiún na Meán’s overall Online Safety Framework, which aims to make digital services legally accountable for how they protect people, especially children, from harm online.

The code will introduce obligations on video-sharing platforms that prohibit uploading or sharing harmful content, including cyberbullying.

It will also prohibit the promotion of self-harm or eating disorders, as well as prohibit incitement to hatred or violence.

The code will also require platforms to use age verification measures to prevent children from encountering pornography or violent videos.

It will also provide parental controls for content which may “impair the physical, mental, or moral development of children under 16”.

Coimisiún na Meán said the code will give it the tools to address the root causes of harm online, including the availability of illegal content, and inadequate protections for children on social media services.

The Online Safety Commissioner, Niamh Hodnett, said the updated Code “is an important step forward to hold platforms to account for keeping people safe online”.

However, the Code has been criticised by CyberSafeKids for not addressing “toxic” algorithms.

The Irish Council for Civil Liberties (ICCL) said it is “disappointed that measures to address toxic algorithms have been removed from the Online Safety Code”, and noted that a previous draft included requirements to turn off recommender systems.

The recommender system is an algorithm that uses data to suggest items that a social media user might be interested in.

An ICCL spokesperson said recommender systems “push hate and extremism into people’s feeds and inject content that glorifies self-harm and suicide into children’s feeds”.

ICCL Senior Fellow Dr Johnny Ryan called it a “dangerous U-turn” and added that social media users should be given the “freedom to decide for themselves whether they can be profiled and fed algorithmic content”.

It does indeed appear that @CNaM_ie has had its wings clipped by @ThierryBreton.
Ireland wanted to force tech firms to switch off toxic recommender algorithms by default (until users decide to switch them on) but was prevented by the European Commission. https://t.co/N9aJSx2Smz https://t.co/RiplS75n91 pic.twitter.com/Xs86JfA5Rx

— Johnny Ryan (@johnnyryan) May 27, 2024

And while CyberSafeKids welcomed the Code described it as a “landmark development”, it also expressed concern at “the fact this code does not address the recommender system”.

A spokesperson for CyberSafeKids, an Irish charity which works to help children, parents and teachers navigate the online world, said a lot of “harmful content coming through a child’s feed originates from this algorithm”.

The spokesperson added that at times, the social media feeds of children and young people are “overwhelmed” by harmful content, and they called for this to be “meaningfully addressed”.

The Code also prohibits platforms from processing children’s data for commercial reasons, but CyberSafeKids said it is concerned that video-sharing platforms “will find ways to circumvent this if possible”.

And when deciding on a user suspension, a video-sharing platform is obliged by the Code to assess cases in a “timely, diligent and objective manner”.

CyberSafeKids said this “lacks the specific parameters we believe are necessary”.

Online campaigning platform Uplift also criticised the Code for not addressing “toxic” algorithms.

Its director Siobhán O’Donoghue said this was an opportunity to “change to the rules for how suggested content is directed at social media users”.

“Turning off recommender systems by default would mean that these profit hungry platforms can’t just decide to flood teenage girls with weight loss content or push porn and gambling content at young men or abusive content targeting election candidates that spreads like wildfire on social media,” said O’Donoghue.

She said a move to “force” video-sharing platforms to “stop automatically sending content based on your personal information, and instead giving users the choice to decide what you, would have been a game changer”.