Nexus Consultation Response:

Volume 3 presents Ofcom’s draft Children’s Register of Risks as part of Ofcom’s duty to assess the risk of harm to children from content online. The draft Guidance on Content Harmful to Children complements the Register of Risks, providing examples of what Ofcom considers to be, or not to be, content harmful to children. 

On Ofcom’s assessment of the causes and impacts of online harms, we would like to see a fuller explanation of the kinds of content harmful to children included in the guidance to support service providers in identifying harmful content. We agree with Ofcom’s analysis of risk factors and content harmful to children, in particular relating to: pornographic content; abuse and hate content; and violent content. We agree that the use of anonymous profiles, private vs open profiles, direct messaging and tagging can increase the risk of children being exposed to harmful content. We would like to highlight the growing use of “Sextortion”, which involves “the threat of sharing images or videos – often ‘nudes’ or sexually explicit content – to extort money or force someone to do something against their will”. Through our work with children and young people, we have also come across a concerning trend using “Live” features on TikTok, Instagram, and Facebook to target children and young people with sexual content. We also included links in our full response to research on the risks of Generative AI to children, including the production and distribution of Child Sexual Abuse Material. 

Nexus agrees with the proposed approach to the draft Guidance on Content Harmful to Children. We welcome Ofcom’s guidance on how content can be highly subjective and context specific, meaning that different kinds of harms can vary in nature due to the presentation of the content and the specific nature of the user and poster. We also welcome Ofcom’s efforts to differentiate between content harmful to children and recovery content, which can be beneficial for children and other users who are on or beginning a recovery journey. 

We agree with the proposal to include codewords, hashtags, substitute terms/phrases, sounds, pornographic GIFs, sexualised emojis, and comments as elements for services to consider as content that poses a risk of harm to children. We would appreciate some clarification around when the use of language strongly associated with sexual activity and pornography is considered to be pornographic content. We have found from our engagement with young people that they don’t often use explicit, sexualised language and are more likely to use acronyms, emojis, and other code words to convey sexualised content. 

In this volume, Ofcom explain their proposals about the governance measures service providers should put in place to manage risk to children and how service providers should go about assessing the risk of harm to children encountering harmful content online. 

We agree with the proposed governance measures. There is a balance of accountability between governance, senior leadership accountability, internal monitoring, and staff policy implementation to create a holistic approach to governance and accountability. We would like to reiterate that assessing risk is often complex and nuanced, and needs supported by other forms of specific training on Child Sexual Exploitation, Child Sexual Abuse, Safeguarding, Child Protection, etc. 

We believe that the proposals in relation to the Children’s Risk Profiles are comprehensive and informative. It is vital that Ofcom creates guidance and assessments that remove any guesswork on behalf of services and directly highlight the harms that children are at risk of coming into contact with and therefore ensures that services completing their Risk Assessments are not self-eliminating their services. We appreciate the emphasis on the detrimental effects of poor governance and accountability on children and vulnerable people, especially if there is a disjointed approach across different codes. 

Volume 5 outlines draft measures Ofcom propose providers of services likely to be accessed by children could take to comply with their child safety duties in the Online Safety Act. These are set out in the draft Children’s Safety Codes in Annexes 7 and 8, which will be finalised following consultation, including age assurance measures, content moderation, search moderation, user reporting and complaints, terms of service, recommender systems, user support, impact assessments, and statutory test. 

We agree with the measures proposed in the Children’s Safety Codes, with some recommendations to add content types relating to sexual abuse and abusive relationships. We also recognise the need for a levelled approach to measures for different services according to risk, size of service, functionalities, and type of service. However, we want to emphasise the need for employing expertise to continually scope for software that might be able to bypass the age assurance technology as this landscape would move quickly. It is vital that the age assurance technology is accurate at determining the age of a user. 

We agree with the proposals on content moderation for user-to-user services. Ofcom’s recommendations include cost-effective, proportionate measures for all services, with extra requirements for large and multi-risk services whereby there is a greater need for protecting service users. We also welcome Ofcom’s commitment to an additional consultation later this year on automated content moderation and detection tools, as we see a growing trend in online technology turning to automated features, AI content and recommendation functions, and automated customer support. We agree with the proposals around search moderation however would like to emphasise the need for services to employ rigorous testing and scoping for any technologies that can enable bypassing or circumventing search moderation and age assurance technologies. 

We agree with the proposed measures for user reporting and complaints. Particularly, we agree that services should create accessible, easy to understand, and transparent complaint and reporting systems that will appeal to children and vulnerable services users and thereby increase the likelihood of harmful content and user profiles being reported. We also agree with the proposed Terms of Service and Publicly Available Statements measures; we believe that language, presentation, and length of the document are key considerations for engaging children, young people, and vulnerable users in the Terms of Service. For example, we have found that employing audio and visual mediums as opposed to text to be more accessible for young people. 

We agree with the proposed recommender system (information filtering through machine learning algorithms) risk mitigation measures. Recommender systems can create a constant stream of harmful content for child users, making it harder for child users to avoid triggering content. Children are at risk of coming across content that they may not wish to see, but because they have engaged with it, they will continue to be recommended said content. 

We agree with proposals around user support and see the importance of these proposals being rolled out to all users to protect them from harmful content. We agree with the proposals around search features, functionalities, and user support, however, recommend adding crisis prevention information relating to sexual abuse and abusive relationships.