What are the inherent challenges of social media? Harm doesn't disappear even with "consent" — We must not conclude social media issues with personal responsibility.

What are the inherent challenges of social media? Harm doesn't disappear even with "consent" — We must not conclude social media issues with personal responsibility.

When we look at our smartphone screens, we feel like we are acting alone.

Watching videos. Liking posts. Skipping over uninteresting information and searching for topics of interest. Each of these actions seems like a highly personal choice.

However, behind the scenes, the behaviors of a vast number of users are collected and analyzed to calculate what to show next to keep us glued to the screen. The impact generated there no longer fits within just an individual's smartphone.

Who is spoken of as the enemy? What is considered "normal"? Which political claims gain momentum? How do young people understand love, sex, and gender relations? The accumulation of recommendations gradually changes the very atmosphere of society.

The real question surrounding social media is not just "Can you delete your data?" or "Did you consent to the use of algorithms?"

The question to ask is how much we should entrust the mechanisms that move the perceptions and emotions of billions of people to companies that aim to maximize advertising revenue.


Three Regulatory Proposals That Started in Australia

In Australia, three policies concerning social media and personal data are currently progressing in parallel.

The first is the "right to erasure." The privacy law amendment proposal announced by the government in August 2026 includes a system that allows users to request the deletion of their personal information from large social media and search services.

The second is a reform that allows users to choose the recommendation functions of social media. Chanel Contos, who leads the anti-sexual violence organization Teach Us Consent, advocates for an "opt-in system" that activates personalization only if the user explicitly wishes for it.

In the initial state, posts from people the user follows are displayed in chronological order. Recommendations based on browsing history and reactions are shown only if the user chooses them.

The third is the "digital duty of care." It aims to impose responsibility on operating companies to reduce foreseeable dangers from the design stage of services, not just deleting problematic posts after damage occurs. The targets may include not only the posts themselves but also recommendation systems, AI, bots, notifications, infinite scrolling, and other functions.

These three may seem similar, but they regulate different targets.

The right to erasure deals with "what companies know about me," while opt-in deals with "what is displayed on my screen." In contrast, the digital duty of care questions "what kind of dangerous services companies should be allowed to provide in the first place."

This last question is the key to advancing the social media issue from individual rights to the structure of society.


Why "Consent" Is Not Enough

It is important to seek consent for obtaining personal information. However, in many digital services, consent has become a formality.

Few people read all the terms of service, understand how data is combined, what inferences it is used for, and which advertisers it affects before consenting. Sometimes, you have no choice but to accept everything if you want to use the service.

Moreover, the social impact of algorithms does not only occur to the person who consented.

If a user is drawn to extreme information, that influence is brought into their family, school, workplace, and community. If someone who believes in misinformation participates in voting or political movements, even those who do not use algorithms cannot escape the results. If misogyny or racism is widely spread, the targeted individuals are affected in their social lives even if they do not use that platform.

We do not think of air pollution as "if you don't want to breathe smoke, just close your windows." Similarly, the pollution of the information environment cannot be solved by each user's settings alone.

Even if you delete your data, attributes and behaviors can be inferred from the data of millions of others. Even if you turn off recommendations for yourself, if people around you are influenced by the same system, social division and radicalization will continue.

Individual rights are necessary, but they are not sufficient on their own.


The Purpose of "Recommendations" Is Not User Happiness

Algorithms themselves are neither good nor bad.

Recommendations hold great value in finding necessary information from a large number of posts, discovering unknown music, hobbies, experts, and support communities. They have also been useful for information transmission during disasters and for minorities to find companions.

The problem is what they are optimized for.

In services that rely mainly on advertising revenue, time spent viewing the screen, clicks, shares, and comments become important. If content that stimulates anger, anxiety, fear, or a sense of superiority gets stronger reactions than calm and convincing information, the system tends to lean in that direction.

Users may not have been looking for extreme ideas from the start. They may enter through general interests like strength training, romance, work, or self-improvement and gradually be led to posts explaining that "a specific group is to blame for my failures."

At this point, companies can claim, "We only displayed information that users showed interest in." However, it is the platform that measures, strengthens, and decides the order of similar information to be shown consecutively.

Was it the user who chose? Or were they guided to keep choosing? The boundary is extremely ambiguous.


Meta's Massive Settlement Indicates a Shift in Responsibility

In August 2026, Meta agreed to a settlement of approximately $18 billion with state and territory attorneys general in the U.S. over lawsuits concerning the harm Facebook and Instagram caused to young people.

The agreement includes usage time limits that young users cannot disable, nighttime usage restrictions, notification suppression during class hours, and parental control features. Payments are planned over ten years, with some conditions requiring YouTube and TikTok to adopt similar standards.

This settlement alone does not establish a causal relationship between social media and all mental harm. However, what is important is that the focus of measures is beginning to shift from "children and parents being careful" to "companies designing products with fewer dangers."

For cars without seatbelts, it is not enough to just say, "Improve your driving skills if you don't want to have an accident." Manufacturers also have a responsibility to enhance safety.

Social media is entering the same stage.


Banning Use for Those Under 16 Was Not the Solution

Since December 2025, major social media platforms in Australia have been required to implement measures to prevent use by those under 16.

After the system started, millions of accounts identified as belonging to minors were deleted, suspended, or restricted. However, it was also reported that some children created new accounts to bypass age verification, and existing accounts remained.

This is not simply a matter of the ban being meaningless. The age restriction has shown the intention to protect children as a society and has pressured companies to take measures.

However, if age alone is the boundary, the moment someone turns 16, they are thrown into an overly personalized feed. Adults are not immune to fraud, misinformation, conspiracy theories, gambling ads, or posts promoting eating disorders.

Instead of just keeping children away from social media, services need to be transformed into safer environments for when children return.


Support for "Default Off" on Social Media, But Concerns Remain

Looking at public social media posts, there is strong support for the idea of disabling algorithms by default.

 

In a Reddit thread discussing Australian news, comments such as "It should be off by default," "I want to see posts from people I follow," and "I only see ads, opinions from strangers, and AI-generated content instead of updates from friends" were common.

Some users, aware of their reactions to posts that provoke anger, welcomed a return to chronological feeds. There were also opinions that viewed the social division caused by algorithms as issues of democracy and national security.

On the other hand, there are opposing or cautious opinions.

In addition to voices emphasizing self-management, such as "If you don't like it, don't use it" and "Education is needed more than regulation," some users do not want to lose the recommendation feeds they have adjusted over a long period. There are concerns about reduced discovery of hobbies, music, local information, and new creators, with suggestions to allow users to choose from multiple display methods like "Following," "Recommended," "Trending," and "Random."

On LinkedIn, it has been pointed out that the binary choice between algorithm or chronological is insufficient. The important thing is what purpose it is optimized for, who manages it, and how much control users have over changes. Researchers commented that while young people value the ability to adjust feeds according to their interests, they are concerned about ads and unrelated posts imposed by companies.

However, these are extracted from public posts and are not public opinion surveys. There is a bias in participants on specific platforms or threads, so it cannot be said that "the majority of the public supports it."

Nevertheless, what becomes apparent from the reactions is that users are not entirely rejecting personalization itself but are strongly dissatisfied with having their emotions and attention exploited in ways they cannot control.


There Are Limits to the Opt-In Approach

If non-personalized display is set as the default, users will be less likely to be incorporated into tracking-based recommendations without being aware of it. This is a clear step forward.

The EU's Digital Services Act also requires large platforms to provide at least one recommendation option not based on profiling.

However, having options and being easy to use are different things. If settings are hidden deep in layers, revert to personalized display when reopening the app, or only apply to part of the functions, the rights become merely formal.

Moreover, even if you turn off algorithms, the extreme accounts you followed until then remain. Harmful information may still arrive through direct messages, searches, shares, comments, and external links. Chronological display is not completely neutral either, as it is influenced by choices like "who you follow" and "which posts are deleted."

While recommendations can sometimes lead people into dangerous "holes," they can also serve as "doors" to new cultures and support.

What is needed is not the elimination of algorithms, but a change in their purpose and power dynamics.


Regulation Should Target Not Just "Content" But "Amplification Mechanisms"

In social media regulation, attention tends to focus on deleting illegal or harmful posts one by one. However, in an environment where a massive number of posts are created every minute, deletion alone is hard to keep up with.

Even the same post has limited impact if it is hardly displayed. Conversely, if it is recommended to millions because it earns reactions, the social impact expands rapidly.

Therefore, what should be demanded of companies is not just the number of deletions.

What kind of posts were spread through recommendations? How long did it take for young people to come into contact with specific harmful information? Did they change the algorithm after risks were pointed out? Did they prioritize safety even if it reduced usage time or ad revenue?

Without allowing supervisory agencies or independent researchers to verify such information, we would have to rely solely on the safety reports published by companies.

An effective system requires not only setting non-personalized display as the default but also risk assessment, external audits, data provision to researchers, explanation of recommendation reasons, obligation to maintain settings, and function tests targeting children.

Furthermore, it is worth considering mechanisms that allow users to choose display methods from multiple feed providers and read post data in different viewing apps. As long as companies monopolize content, relationships, and display rankings as a whole, users' "choices" remain limited.


It's Not Just Someone Else's Problem for Japan

The discussions in Australia are directly relevant to Japan as well.

In Japan, issues like defamation, misinformation, investment fraud, recruitment for illegal jobs, sexual image damage, and information manipulation during elections on social media are problematic. However, countermeasures tend to focus on individual units, such as punishing posters, disclosure requests by victims, and user warnings.

Of course, holding malicious posters accountable is essential. However, if the design that prioritizes the widespread dissemination of provocative posts and connects perpetrators with potential victims is left unchecked, the same problems will repeat.

Measures like "Let's not be deceived," "Block if you don't want to see it," and "Teach rules at home for children" alone cannot bridge the power gap between the world's largest technology companies and individual users.

A system is needed that does not place all responsibility on education, alongside digital education.


The Question Should Be "Whose Interests Does the Technology Serve?"

The right to erasure, consent, opt-out, and opt-in are all important tools. However, they are starting points, not ultimate solutions.

A system that assumes all users perfectly manage their settings, discern suspicious information, not react to emotionally stimulating posts, and constantly monitor children's usage is unlikely to function well in reality.

There is a more fundamental question to ask.

Is an advertising system based on large-scale surveillance and behavior prediction truly necessary for society? Can the success of recommendation systems be evaluated by metrics other than usage time and reaction numbers? Should services with influence comparable to public spaces bear responsibility for public interest?

The fact that an individual once clicked "agree" cannot justify all subsequent social impacts.

The social media screen appears private. However, the recommendations repeated billions of times there create the reality we share.

Therefore, the necessary question is not just "Do I want to use this algorithm?"

"What kind of information environment do we want to live in together?"

The discussion on social media regulation is finally beginning to approach that question.


Sources

Related Articles