1. Introduction

Civil Law and Social Media Algorithm Mental Health Harm Claims in Singapore concerns legal disputes arising when social media platforms, recommendation algorithms, automated content-ranking systems and personalised digital advertising allegedly contribute to psychological or psychiatric harm.

Social media platforms commonly use algorithmic systems to determine:

Which posts appear in a user's feed.

How frequently particular content is recommended.

Which videos and images receive greater visibility.

Whether users are exposed to emotionally disturbing material.

How advertisements are personalised.

Which notifications encourage repeated platform engagement.

How user behaviour influences future recommendations.

These technologies may raise civil liability questions where their design, operation or management allegedly contributes to depression, anxiety, psychological distress, compulsive platform use, eating disorders, sleep disruption or other recognised psychiatric injuries.

The central legal question is:

Can a social media platform, algorithm developer, advertiser or other responsible party be held legally liable when its algorithmic design or content-recommendation practices cause or materially contribute to a user's mental health injury?

Singapore does not currently have a dedicated statutory tort specifically addressing mental health injuries caused by social media recommendation algorithms. Such disputes would therefore need to be examined through existing principles of negligence, duty of care, contract, consumer protection, privacy, defamation and applicable statutory regulation.

2. Legal framework governing social media algorithm mental health claims

A. Constitution of Singapore

The Constitution provides the broader framework for individual rights and state responsibilities.

Relevant provisions include:

These provisions provide important constitutional context, but they do not automatically create a private damages claim against a social media company for algorithm-related psychological harm.

B. Personal Data Protection Act 2012 (PDPA)

The PDPA is relevant where social media organisations collect, use or disclose personal information.

Potential disputes include:

Collection of behavioural information without appropriate consent or another applicable legal basis.

Excessive collection of personal information.

Improper use of psychological or behavioural profiles.

Unauthorised disclosure of sensitive personal information.

Failure to protect personal data.

Inaccurate personal information used in automated recommendation systems.

For example, a platform might use a user's browsing patterns, engagement history and interactions to personalise content.

If the organisation breaches an applicable PDPA obligation, regulatory consequences and statutory remedies may arise, subject to the relevant provisions.

Importantly, the PDPA does not automatically provide compensation for every instance of psychological harm associated with personalised advertising.

C. Protection from Harassment Act 2014

The Protection from Harassment Act (POHA) may become relevant when algorithmic content distribution involves harassment, threatening communications, stalking or other conduct covered by the Act.

Potential situations include:

Repeated targeted harassment.

Coordinated online abuse.

Distribution of threatening content.

Repeated publication of harmful communications.

Online conduct causing legally recognised harassment-related harm.

A recommendation algorithm merely displaying controversial content does not automatically constitute harassment. The statutory elements must be established.

D. Consumer Protection (Fair Trading) Act 2003

The Consumer Protection (Fair Trading) Act (CPFTA) may be relevant where a platform engages in unfair practices in consumer transactions.

Potential issues include:

Misleading representations about platform safety.

Misrepresentations concerning parental controls.

Unfair subscription practices.

Misleading descriptions of privacy protections.

Material omissions in consumer-facing representations.

A mental health injury alone does not establish a CPFTA claim. The claimant must demonstrate an applicable unfair practice and satisfy the statutory requirements.

E. Online Safety and broadcasting regulation

Singapore's Broadcasting Act and applicable IMDA regulations provide a regulatory framework relevant to online content services.

Online safety obligations may address harmful content, platform safeguards and regulatory directions.

These regulatory requirements should be distinguished from private civil causes of action. A regulatory breach does not necessarily establish an individual entitlement to damages.

3. Major categories of mental health harm caused by social media algorithms

3.1 Algorithmic amplification of harmful content

Recommendation algorithms may repeatedly display content involving:

Self-harm and suicide.

Violence and traumatic events.

Body-image dissatisfaction.

Eating-disorder promotion.

Humiliation and bullying.

Extreme or disturbing material.

A claimant may argue that a platform's recommendation system repeatedly amplified harmful material despite indicators that the user was vulnerable.

The legal question is whether the platform's conduct breached an independently established duty or statutory obligation and whether that breach caused a recognised injury.

3.2 Compulsive engagement and addictive design

Social media platforms may use:

Infinite scrolling.

Continuous autoplay.

Personalised notifications.

Variable content rewards.

Engagement-based recommendations.

Repeated prompts to return to the application.

Potential allegations include negligent design, misleading consumer representations and failure to comply with applicable regulatory obligations.

However, prolonged use or compulsive engagement does not automatically establish a legally recognised psychiatric illness or negligence.

3.3 Body-image and eating-disorder harm

Algorithmic recommendation systems may repeatedly expose users to:

Unrealistic beauty standards.

Extreme dieting content.

Weight-loss advertisements.

Body-comparison material.

Potentially harmful fitness recommendations.

A claimant may allege that targeted recommendations contributed to an eating disorder, depression or anxiety disorder.

Relevant evidence would include the user's exposure history, medical records, psychiatric assessments and expert evidence concerning causation.

3.4 Cyberbullying and algorithmic distribution

Algorithms may increase the visibility of humiliating, threatening or abusive content.

Potential claims involve:

Repeated harassment.

Coordinated online attacks.

Viral distribution of humiliating material.

Repeated exposure to threatening communications.

Failure to address notified online abuse.

The platform's response after receiving a sufficiently detailed complaint may be especially important.

3.5 Personalised advertising and psychological exploitation

Behavioural advertising may use engagement and browsing information to target individuals.

Potential disputes include:

Exploitation of behavioural vulnerabilities.

Misuse of personal information.

Misleading mental-health product advertisements.

Inappropriate targeting of minors.

Repeated advertising of harmful products.

The claimant must distinguish unlawful data processing or misleading commercial conduct from the mere fact that an advertisement was personalised.

3.6 Mental health harm involving children and teenagers

Children may have particular vulnerabilities relating to:

Social comparison.

Peer approval.

Cyberbullying.

Body-image concerns.

Exposure to harmful material.

Excessive platform engagement.

Where children are involved, the platform's knowledge, available safeguards, representations to parents and applicable regulatory requirements may become important in assessing liability.

4. Important Singapore case laws

The following six authorities provide relevant legal principles. There is currently no established Singapore appellate judgment specifically awarding damages against a social media company for psychiatric injury caused by its recommendation algorithm. Accordingly, these authorities must be understood as direct psychiatric-harm authorities or supporting analogies.

Case 1. Spandeck Engineering (S) Pte Ltd v Defence Science & Technology Agency [2007] 4 SLR(R) 100

Leading negligence authority

Area: Duty of care and negligence.

Facts: The dispute arose from construction-related defects and alleged negligence involving a government agency.

Decision: The Court of Appeal established Singapore's general two-stage framework for determining whether a duty of care exists.

The framework requires consideration of:

Factual foreseeability and legal proximity.

Policy considerations that may negate the imposition of a duty.

Application to social media algorithms:

A claimant may argue that a platform knew or ought reasonably to have known that a particular recommendation design created foreseeable risks of psychiatric injury.

For example:

Repeated amplification of self-harm content.

Personalised recommendations involving eating disorders.

Deliberate targeting of vulnerable users.

Failure to implement reasonable safety controls.

However, foreseeability alone does not establish liability.

The claimant must demonstrate sufficient legal proximity and overcome relevant policy considerations.

Legal principle: A social media platform's duty of care cannot simply be assumed because it operates a recommendation algorithm. It must be established through the applicable negligence framework.

Case 2. Pang Koi Fa v Lim Djoe Phing [1993] 3 SLR(R) 73

Area: Psychiatric injury and negligence.

Facts: The litigation concerned psychiatric harm arising in a medical context following a patient's death.

Decision: The High Court considered the principles governing recovery for psychiatric injury, including the importance of the relationship between the parties and the circumstances in which the injury occurred.

Application to social media algorithms:

This case provides an early Singapore foundation for determining when psychiatric injury is legally recoverable.

In an algorithm-related claim, the court may need to examine:

The relationship between the user and platform.

The nature of the alleged exposure.

Whether psychiatric injury was reasonably foreseeable.

The circumstances connecting the defendant's conduct to the injury.

Legal principle: Psychiatric injury claims require more than proof of emotional upset; the legally relevant psychiatric harm and relationship between the parties must be established.

Case 3. Ngiam Kong Seng and another v Lim Chiew Hock [2008] 3 SLR(R) 674; [2008] SGCA 23

Leading psychiatric injury authority

Area: Recognised psychiatric illness and duty of care.

Facts: A serious traffic accident left one claimant with severe physical injuries. His wife subsequently claimed clinical depression arising from the consequences of the accident and her caregiving responsibilities.

Decision: The Court of Appeal clarified how psychiatric-harm claims must be analysed under the Spandeck framework.

The Court emphasised two threshold requirements:

A recognised psychiatric illness.

Factual foreseeability of psychiatric harm.

The court then considers legal proximity and public-policy factors.

Application to social media algorithms:

This judgment is particularly relevant to claims involving:

Depression allegedly caused by repeated algorithmic exposure.

Anxiety disorders.

Psychiatric injury arising from cyberbullying.

Psychological injury associated with repeated harmful recommendations.

A claimant should provide reliable psychiatric evidence rather than relying solely on statements of stress, sadness or frustration.

Legal principle: Pure mental harm caused by negligence is not automatically compensable. A recognised psychiatric illness and the other requirements for negligence must be established.

eLitigation+1

Case 4. Man Mohan Singh s/o Jothirambal Singh and another v Zurich Insurance (Singapore) Pte Ltd and another [2008] SGCA 24

Area: Depression, post-traumatic stress and recoverability of damages.

Facts: The claimants were parents whose two teenage sons died in a road accident. They sought damages including compensation for post-traumatic shock and depression.

Decision: The Court of Appeal examined the psychiatric injury claims and the distinction between compensable psychiatric illness and ordinary grief.

The judgment also considered the limits of negligence liability and the requirements of legal proximity and foreseeability.

Application to social media algorithms:

A user claiming that algorithmic recommendations caused depression must distinguish between:

Temporary emotional discomfort.

Ordinary disappointment or distress.

A clinically recognised psychiatric disorder.

For example, a user alleging depression following repeated exposure to harmful body-image content would need supporting medical and psychiatric evidence.

Legal principle: Severe emotional suffering does not necessarily establish a compensable psychiatric illness.

Case 5. Li See Kit Lawrence v Debate Association (Singapore) [2023] SGHC 154

Important modern psychiatric-harm authority

Area: Online publication, psychiatric injury, negligence and intentional mental harm.

Facts: The case concerned public statements published on a website, Facebook page and Facebook group regarding allegations of misconduct against a former member of an organisation.

The deceased had previously experienced mental health difficulties. His father alleged that the organisation's disciplinary actions and public communications caused an acute stress reaction that led to his son's suicide.

Decision: The High Court granted declarations that certain disciplinary decisions were unlawful and set them aside because they were ultra vires and breached natural justice.

However, it dismissed the negligence and other tortious claims. The plaintiff had not sufficiently proved that the deceased suffered the alleged recognised psychiatric illness immediately before his death or that the necessary causal connection had been established.

Application to social media algorithms:

This is particularly important for the proposed subject because it involved online publication through Facebook.

It demonstrates the need to prove:

The specific harmful communication.

The defendant's relevant legal duty.

A recognised psychiatric illness.

Medical causation.

The connection between the publication and the injury.

It also illustrates that a platform-related psychological harm claim cannot succeed merely because an individual experienced serious distress after online publication.

Legal principle: Psychiatric injury and causation require reliable evidence. The court will distinguish severe distress from a proven psychiatric illness.

eLitigation+2

Case 6. Miya Manik v Public Prosecutor [2021] 2 SLR 1169

Area: Medical expert evidence and psychiatric diagnosis.

Facts: The case involved the evaluation of psychiatric reports tendered in criminal proceedings.

Decision: The Court of Appeal considered the importance of properly substantiated expert evidence. It emphasised that experts should explain the evidence and reasoning underlying their conclusions rather than simply announcing a diagnosis.

Application to social media algorithms:

This case is useful when a claimant relies on psychiatric reports to establish algorithm-related mental injury.

For example, a report should explain:

The diagnostic criteria applied.

The clinical symptoms identified.

The relevant medical history.

The timing of the psychiatric illness.

The possible alternative causes.

The reasoning connecting the alleged exposure with the condition.

A bare assertion that a social media algorithm caused depression may be insufficient.

Legal principle: Expert psychiatric evidence must be reasoned, adequately supported and capable of judicial evaluation.

5. The Online Safety (Relief and Accountability) Act 2025

Important legislation effective 29 June 2026

An important development in Singapore's online-harm framework is the Online Safety (Relief and Accountability) Act 2025 (OSRAA).

Its statutory tort provisions commenced on 29 June 2026. The legislation creates legal duties for three categories of online actors:

Communicators.

Administrators.

Online platforms.

The legislation provides a mechanism for victims of specified online harms to pursue civil claims against responsible parties. Remedies may include damages and orders to stop or address the harmful conduct.

Ministry of Law - Online Safety+1

A. Platform responsibility

An online platform must respond reasonably after receiving a qualifying online harm notice.

Its responsibilities may include:

Assessing whether the reported harmful activity occurred.

Taking reasonable measures to address the harm.

Preventing continuing harmful activity.

Responding within a reasonable time.

The reasonableness of the response depends on the circumstances, severity and likelihood of continuing harm.

Ministry of Law - Online Safety

B. Application to algorithmic mental health claims

Consider a hypothetical situation:

A teenager repeatedly receives algorithmically recommended self-harm material. The teenager also experiences targeted harassment through the platform. A parent reports the harassment through the prescribed online harm notification procedure.

Potential questions include:

Did the reported activity fall within a category covered by OSRAA?

Did the platform receive a legally sufficient notice?

Did it respond reasonably?

Did the platform take adequate measures to address continuing harmful activity?

Is the claimed psychiatric injury legally recoverable?

Important limitation: OSRAA does not make every harmful recommendation or mental health condition an independent statutory tort. The particular conduct must fall within the statutory categories and requirements applicable at the relevant time.

Generic depression, anxiety or compulsive social media use is not automatically a covered statutory online harm.

Ministry of Law - Online Safety+1

6. Establishing civil liability against social media companies

A claimant may consider several legal routes.

A. Negligence

The claimant must establish:

A duty of care owed by the defendant.

Breach of the applicable standard of care.

Factual and legal causation.

Recognised and legally recoverable damage.

Potential allegations include inadequate safety testing, negligent recommendation-system design, failure to address known risks and inadequate protection of vulnerable users.

The Spandeck framework will be central to establishing duty of care.

B. Breach of contract

A platform may make representations or contractual commitments concerning:

User safety.

Privacy protection.

Content moderation.

Parental controls.

Account protection.

Subscription services.

If the platform breaches an enforceable contractual obligation, the user may pursue contractual remedies.

However, damages for mental distress are not generally available for every contractual breach.

C. Statutory online-harm claims

Where the conduct falls within OSRAA or POHA, the claimant may pursue applicable statutory remedies.

These laws are especially relevant where algorithmic distribution interacts with harassment, stalking, doxxing or other specified online harms.

D. Data protection claims

A claimant may investigate whether the platform improperly collected, used or disclosed personal information.

For example, a dispute could involve behavioural profiling, unauthorised use of personal information or failure to protect personal data.

E. Defamation and malicious falsehood

Where algorithms distribute false and damaging statements about an individual, defamation or malicious falsehood claims may arise if their respective legal requirements are satisfied.

Algorithmic amplification alone does not establish that a platform is legally responsible for every statement distributed by its users.

7. Parties potentially responsible for algorithm-related mental health harm

Liability must be determined separately for each defendant. A platform's responsibility cannot automatically be transferred to its software developer or advertising partner.

8. Evidence required in mental health litigation

Evidence is essential because proving that a user encountered harmful content is different from proving that the content caused a psychiatric illness.

Algorithmic causation

The claimant may need to establish the relationship between:

The recommendation system.

The content repeatedly delivered.

The user's exposure.

The development or worsening of the psychiatric condition.

Alternative explanations must also be examined, including pre-existing psychiatric conditions, family circumstances, workplace stress and other independent events.

9. Available civil remedies

Depending on the cause of action and statutory requirements, possible remedies include:

A claimant should not assume that every psychiatric injury will result in an award of damages. The court must consider liability, causation, remoteness and the relevant legal requirements.

10. Future legal reforms in Singapore

Several developments could improve accountability in this area.

A. Algorithmic transparency

Platforms could be required to provide meaningful explanations of recommendation practices and safety-related system decisions.

B. Independent algorithmic audits

Independent reviewers could examine whether recommendation systems repeatedly expose vulnerable users to harmful content.

C. Child-protection mechanisms

Platforms could strengthen age-appropriate recommendations, parental controls and safeguards against harmful content.

D. Mental health risk assessments

Platforms could conduct systematic assessments of foreseeable psychological risks associated with major recommendation-system changes.

E. Stronger complaint mechanisms

Users should have accessible methods of reporting harmful algorithmic recommendations and obtaining meaningful responses.

F. Preservation of digital evidence

Platforms should maintain appropriate records that allow courts and independent experts to investigate algorithmic behaviour and causation.

11. Consolidated case-law reference table

12. Critical legal analysis

Three major legal questions are likely to determine the development of this subject.

Question 1: Can a social media platform owe a duty of care to users against algorithm-related psychiatric injury?

Potentially, but this remains an issue requiring development through Singapore case law. Under Spandeck, factual foreseeability, proximity and policy considerations must be examined. The nature of the platform-user relationship and the platform's knowledge of specific risks may be relevant.

Question 2: Is repeated exposure to harmful content sufficient to establish psychiatric injury?

No. Repeated exposure may establish part of the factual background, but a negligence claim for pure mental harm generally requires proof of a recognised psychiatric illness, together with the other elements of liability.

Question 3: Can algorithmic recommendation systems themselves become the subject of civil litigation?

Yes, potentially. Their design, testing, operation, personalisation and response to known risks may be relevant evidence. However, the mere existence of an algorithm or an adverse psychological outcome does not automatically establish liability.

13. Conclusion

Civil Law and Social Media Algorithm Mental Health Harm Claims in Singapore represents an emerging area connecting negligence, psychiatric injury, online safety, consumer protection, privacy and artificial intelligence governance.

The most important legal foundations are:

The Spandeck duty-of-care framework.

The recognised psychiatric illness requirement in Ngiam Kong Seng.

The psychiatric injury principles in Man Mohan Singh.

The online publication and causation analysis in Li See Kit Lawrence.

The importance of reliable medical evidence in Miya Manik.

The new statutory remedies under OSRAA.

Final legal proposition: Social media companies should not automatically escape legal scrutiny merely because harmful outcomes arise through automated recommendation systems. Equally, they should not automatically become liable for every psychological difficulty experienced by their users. Civil liability requires proof of an applicable legal duty, breach, causation and legally recognised harm.

Singapore's developing online-safety legislation provides additional mechanisms for addressing specified online harms, while future judicial decisions will determine how far traditional negligence principles extend to algorithmically induced psychiatric injury.

Research qualification: The six cases above are Singapore authorities supporting the relevant legal principles, not six judgments directly deciding social media algorithm mental health liability. No external reading links have been included.