Will social media bans reduce the amount of substance use‐related content to which young people are exposed?

Australia recently implemented a social media ban to reduce young people's exposure to content that could harm their wellbeing, like alcohol and substance use-related content. This editorial discusses the potential impact that social media bans may have on exposure to substance-related content and notes that the effectiveness of the ban may depend on how young people and industry respond to the ban. It also identifies what other policy changes could be effective to make social media platforms safer. Social media sites have become extremely popular, particularly among young people [1]. However, there are ongoing concerns about the type of content young people see on social media sites and the potential impact on their wellbeing. These concerns have led to proposals to ban young people from social media. For example, Australia recently implemented a ban to prohibit those under 16 years old from creating or keeping accounts on certain social media sites (e.g. Instagram, TikTok) by requiring age verification, and a number of other countries are considering similar legislation [2]. To date, the vast majority of attention and debate has centred on whether a social media ban will be effective for improving young people's mental health [3]. However, the ban was also driven by concerns about exposure to adult-related content (including alcohol and other substances) [2]. In this editorial, we consider what impact a social media ban will have on young people's exposure to substance-related content and identify that how young people and industry (e.g. social media, alcohol and tobacco) respond to the ban may impact exposure. Additionally, we overview other potential strategies that, as part of a cohesive policy package, could be implemented to make social media sites safer for young people. Recent systematic reviews and meta-analyses suggest that young people who spend more time on social media or report greater exposure to substance-related content also report greater substance use (alcohol [4-6], vaping [7] and multiple substances [8]). Groups like the World Health Organisation (WHO) recommend advertising bans as one of their ‘best buy’ policies, and alcohol advertising is common on social media sites. Indeed, even those who cannot legally purchase alcohol report seeing alcohol advertising [9]. The most commonly cited mechanism for the link between social media use or exposure and alcohol use is social norms theory [6], which suggests that exposure to substance-related content leads to substance use through promoting the belief that substance use is normal and accepted [6]. Evidence suggests that greater exposure during adolescence heightens normative beliefs that give rise to more substance use in young adulthood [10]. Therefore, social media bans may help to reduce the amount of substance-related content that young people see, which could have downstream effects on substance use rates later in life. The impact of the ban on exposure to substance-related content will likely depend on: (1) the effectiveness of the ban at keeping young people off social media (e.g. whether young people stay off social media sites or if they circumvent the ban by using adult accounts or VPNs); and (2) what activities replace social media site use (e.g. whether young people replace it with in-person activities or with other media activities/sites not yet covered by the ban). We hypothesise that young people who circumvent the ban by registering as adults may experience an increase in the amount of substance-related content they see on social media. Currently, platforms like Instagram use ‘teen accounts’ to restrict certain sensitive content to younger people, including vaping, other drug-related content and alcohol advertising. Underage users who access the sites as adults will, therefore, lose these protections. However, we note that this increase in exposure may differ depending on the type of substance. For example, although those who circumvent the ban may see more vaping, smoking or other drug content, sensitive content does not appear to include user-generated alcohol content. Indeed, there is evidence that user-generated alcohol posts (i.e. posts from friends that include alcohol) and non-traditional alcohol advertising (like influencer marketing) are still visible with teen accounts [9]. For young people who adhere to the ban, we believe that the impact on exposure will depend on the ‘displacement effect’ [11]. If social media time is replaced by offline activities, exposure may decrease. However, if the time spent on social media is replaced with time spent on other digital media, such as gaming, streaming or other online activities, the outcome is less certain. There is a risk of migration to ‘darker’ platforms not included in the social media ban (e.g. 4chan) with negligible content moderation [12]. More importantly, the focus on ‘banning’ users rather than ‘fixing’ platforms ignores safety-by-design principles [13]. If the underlying algorithms of alternative platforms continue to prioritise high-arousal substance content to drive ‘watch time’, then removing one platform may simply redistribute exposure rather than eliminating it. Effective policy must address the inherent safety of the digital architecture itself, rather than relying solely on the exclusion of specific age groups. At this stage, it is difficult to determine the impact of a ban on exposure to substance-related content, but we believe that the ban provides an excellent opportunity for research. First, to understand the impact on exposure, research must prioritise evaluating how young people respond to the ban and how many circumvent it. Doing so will allow us to understand any potential iatrogenic effects of the ban, specifically focusing on the impact of exposure for those who bypass age gates and use the platforms as adults. This entails also testing how effective the ‘sensitive content’ filters are at reducing substance-related content and how large the increase in exposure is (which may vary by substance). Further, research should also prioritise identifying where young people who adhere to the ban go and what impact this has on exposure. This may also be particularly important and helpful to understand how industry responds and if they pivot to advertise on platforms that young people move to. In the past, when traditional media channels were restricted, tobacco and alcohol industries [14] pivoted to ‘below-the-line’ marketing (e.g. sponsored posts by influencers) to maintain brand saliency among youth [9]. In the context of a social media ban, we expect an intensification of alcohol references in other media that allow advertising, such as music and movies [15, 16], and even generative AI platforms like ChatGPT (who recently announced advertising). We suggest that evaluation designs for this policy must look beyond platform-specific metrics and adopt a cross-platform lens to capture the ‘whack-a-mole’ effect of marketing displacement, ensuring that reductions in social media exposure are not simply offset by increased saturation in other digital third spaces. One benefit of the ban is that it has shown that there is political will to make social media sites safer for vulnerable populations, and this may be an opportunity to push for more comprehensive, safety-by-design policies. Rather than reactive moderation, platforms should be required to perform mandatory risk assessments to ensure their engagement-driven algorithms do not inherently prioritise substance-related content. Because the Australia ban is limited to those under 16 years, a broader ‘safety by default’ approach is needed to protect all users. This should include a total ban on alcohol advertising, including ‘below-the-line’ strategies that may be supported by strict transparency requirements [9] and the inclusion of alcohol in the definition of ‘sensitive content’. Additional policy could push to mandate platforms to disclose the performance of their algorithms used to filter ‘sensitive content’ from people's feeds. By shifting the burden from the user to the system's architecture, we can ensure that high-accuracy detection algorithms are used as safeguards to proactively de-prioritise user-generated substance imagery. At the individual level, schools could implement media literacy programs [17], which have been shown to be effective at reducing substance use. These programs help young people learn how to recognise manipulative advertising, identify misinformation and critically evaluate online content related to alcohol, vaping and other substances [18]. Programs could also include information about how to turn on settings to remove ‘sensitive content’ or remove alcohol advertising from their feeds and tools like alcohol browser plugins, where algorithms can be incorporated to scan images and block alcohol and substance-related content (currently under development and qualitative evidence cites strong support, especially for use with minors) [19]. It is difficult to determine what impact a ban alone will have on exposure to substance-related content, and it will likely depend on whether young people circumvent the ban, what replaces time spent on social media and how industry responds. Additional policy changes could work synergistically with a ban to reduce substance-related content online, which would benefit all users. Ultimately, although a ban is a significant political statement, true harm reduction may require a shift from policing user age to regulating the safety and transparency of the digital environments themselves. Benjamin Riordan: Conceptualization (lead); writing—original draft (equal). Megan Strowger: Writing—review and editing (equal). Emmanuel Kuntsche: Writing—review and editing (equal). Taylor Winter: Writing—original draft (supporting); writing—review and editing (equal). Benjamin Riordan was sponsored by an ARC DECRA Fellowship. None.