Introduction
The Internet began as a small, restricted virtual network for real time communication within universities and research institutes. It has since dramatically evolved. To build a sense of community within the many social spaces of this early Internet, users often created the rules and guidelines for participation, standards of behaviour and other forms of self-management and control, appointing ‘superusers’ to enforce them[1]. When the Internet opened to the public, it was perceived as a communication architecture serving freedom of expression. Technological advances fostered a democratisation of online speech, while mainstreaming the exchange of digital content. The decentralised nature of the Internet complicates censorship efforts by governments and companies, unlike in the mass media[2]. Anyone connected can produce, disseminate and access information on the network. Nevertheless, since their inception and throughout the past decades, online social spaces have established and enforced rules of conduct to face the infinite growth of the internet and the contents produced.
Content moderation on social media and news platforms is therefore not new. What is new is that content moderation is now organised on an industrial scale, by professionals paid for their assessment and monitoring services. This activity is carried out on behalf of commercial entities: social media platforms, news agencies, companies whose online presence requires monitoring, dating apps and websites, etc. This phenomenon has developed on a large scale, keeping pace with the proliferation of social media, the search for digital information, social interactions and other online activities that are now part of our daily lives. Due to the incredible reach and impact of mainstream social media, these companies need a global workforce available around the clock to monitor contents, protect their brand image and comply with States regulations.
In her book Behind the Screen: Content Moderation in the Shadows of Social Media[3], Sarah T. Roberts describes the toxic and alienating nature of this job, as well as the major role played by online content reviewers in today’s world. The public has become increasingly concerned about the reliability of social media contents and the information it circulates. In 2017, content moderation became a hot topic following the broadcast, sometimes live, on Facebook and other social media platforms, of a series of tragic, violent and highly publicised events. These events prompted public debate regarding what circulates online and how, and who, if anyone, is responsible for monitoring it[4].
However, much of the content moderation on the Internet is carried out not using sophisticated artificial intelligence and machine learning algorithms, but by low-paid workers who, due to the nature of their work, are exposed to psychosocial risks leading to burnout, depression, apathy, and worse still. Through leaks to the press, online content reviewers for Facebook, YouTube and other major platforms have spoken out and raised awareness about their role and working conditions. Similarly, the limits and impact of social media and algorithms have been studied by computer scientists and social science researchers[5].
In the 2010s, online content moderation became industrialised, toxic and alienating[6]. This work was organised in large-scale, outsourced, industrialised offices, similar to call centres and often located in countries where labour is plentiful and cheap (‘crowd working’). For example, thousands of people in Colombia moderate content for TikTok. Hired by a third party, these workers report illegal and inappropriate content on the platform on a daily basis, including cannibalism, pornography, paedophilia and murders. However, many workers report receiving no psychological support, despite demanding or impossible-to-meet performance targets and facing punitive wage deductions and extensive surveillance during their working hours[7]. Most labelling tasks are carried out by workers in low paid regions of the world such as East Africa, India, or the Philippines[8].
Online content reviewers are exposed to significant risks of psychological distress, secondary traumatic stress, anxiety, depression, sleep disorders, emotional exhaustion, cynicism, desensitisation and detachment[9]. Symptoms comparable to post-traumatic stress disorder (PTSD) are frequently observed, particularly following repeated exposure to violent, hateful or sexually explicit content, such as child sexual abuse material[10]. The cumulative effects of this exposure can lead to burnout and have repercussions on personal life and social relationships[11]. Milagros Miceli compares this current way of working with coal mining: mining exposes workers to lung diseases and poses a threat to physical health, while online content moderation exposes workers to ‘toxic and disturbing’ content that poses a threat to mental health[12]. In such settings, the combination of traumatic work content, algorithmic management, and precarious working conditions can cause a combination of OSH risks posing a significant threat to worker’s health[13].
Role and characteristics of online content reviewers
The role of online reviewers is to examine and filter user-generated content published on websites, social media platforms and online marketplaces. This review may take place before the content is published and integrated into the site, or after it has already been posted online. It is typically carried out in response to complaints raised by site administrators, external parties (e.g., companies alleging the unlawful appropriation of content they own) or by users who are disturbed or concerned and who trigger an action known as ‘flagging’ content for human review[14].
Content moderation is a core part of how commercial websites, social media platforms and other digital distribution channels operate. These platforms rely on user-generated content to build their audiences, and moderation is how they manage what gets published (or unpublished). Moderation and filtering are crucial steps that protect the brand image of the company or platforms by ensuring users comply with the site’s guidelines or terms of use, guaranteeing the enforcement of laws and regulations governing their activities, and helping to maintain an audience of users keen to publish and view content on their sites. According to Recital 45 of the EU Digital Services Act[15], providers “should clearly indicate and maintain up to date in their terms and conditions the information as to the grounds on the basis of which they may restrict the provision of their services. In particular, they should include information on any policies, procedures, measures and tools used for the purpose of content moderation, including algorithmic decision-making and human review, as well as the rules of procedure of their internal complaint-handling system”[16].
However, on many high-traffic websites, the volume of user-generated content becomes so large that the sorting and moderation process is being called into question. Sorting content into the ‘acceptable’ or ‘rejected’ piles far exceeded the capabilities of software or algorithms alone in 2019[17]. However, capacities improve as algorithms are continuously trained by human content review. The engagement with digital technologies and platforms by users is driven by a constellation of structural, organisational, and individual-level factors[18].
Online content reviewers are today most of the time precarious platform workers and rarely employed by websites or platforms directly. This process of outsourcing tasks and creating so-called microwork is even sometimes thought by developers as ‘a framework that allows one to think of the people as software components’[19]. In addition, online content reviewers may be employees, self-employed, platform workers, or volunteers, with very different levels of protection and support[20].
The challenges of estimating the number of online content reviewers
There are currently no robust, official statistics that directly estimate the percentage of workers performing content review either in the EU or globally. Labour force and enterprise surveys in the EU and internationally do not contain a distinct occupational or industry code for online content reviewers and there is no official Eurostat, ILO, or OECD figure for the share of workers whose main task is content review, either within the EU or globally. In addition, the outsourcing through complex subcontracting chains further obscures any attempt to calculate workforce shares[21].
Research and policy documents stress the opacity of the sector and the reluctance of large platforms and outsourcing providers to disclose headcounts and employment structures, so any numerical approximation must be treated with caution[22]. Human‑rights and investigative reports note that “a veil of secrecy surrounds this industry” and that “no major social media platform has been willing to say how many online content reviewers are hired on behalf of outsourcing firms”.
Empirical work on content moderation in large online platforms (meta‑platforms and ‘very large online platforms’ (VLOPs)) highlights a large number of review teams, fragmented across languages, locations, and contractual arrangements. In content moderation for VLOPs in the EU, strong disparities in moderator–to–user and moderator–to–language ratios were found across platforms. Transparency reports from platforms reveal that moderation decisions are now overwhelmingly automated (more than 90% on Facebook and Instagram are made by AI systems). However, independent investigations emphasise that despite growing automation, the absolute number of moderators remains large and increasing in response to escalating volumes of user‑generated content and regulatory pressures[23]. Case studies and journalistic investigations suggest that large platforms individually employ, directly and indirectly via subcontractors, workforces ranging from several thousand to low tens of thousands worldwide, dispersed across the EU, North America, Latin America, Africa, and Asia[24].
Academic and policy publications call for enhanced transparency obligations on platforms and their contractors, including public reporting of the number, location, contractual status, and language coverage of online content reviewers, which would enable more precise estimates of their share in the labour force[25]. Their share in total employment appears small, but their societal importance is disproportionately high[26].
Effects of online content reviewing on online content reviewers’ mental health and exposure to PSR factors
Online content reviewing poses significant psychosocial risks (PSRs) and has adverse effects on mental health. The risks arise from the nature of the content reviewed, the use of algorithmic management systems, and often precarious employment and working conditions. These risk factors compound, creating cumulative strain and vulnerability to stress‑related and trauma‑related outcomes in this workforce. The more frequently reviewers are exposed to disturbing content, the greater the risk of psychological distress and secondary trauma[27].
Mental health risks from exposure to work content
Qualitative and quantitative studies show that sustained exposure to disturbing material, such as child sexual abuse materials, graphic violence, hate speech, and self‑harm content, can lead to a spectrum of psychological harms among online content reviewers. Reported outcomes include elevated anxiety, depressive symptoms, sleep disturbance, intrusive thoughts, and in some cases symptoms consistent with post‑traumatic stress disorder (PTSD). A study of 160 online content reviewers working with extreme material found that more than a quarter exhibited moderate to severe psychological distress, although access to a dedicated wellbeing service mitigated some of these effects[28]. Qualitative interviews identify themes such as desensitisation, hypervigilance around children, viewing the world as a ‘darker place’, and strain on family relationships, indicating that harms extend beyond the workplace into broader cognitive and emotional functioning[29].
In psychosocial risk terms, this work involves high emotional demands, the risk of secondary or vicarious trauma, and moral distress. The distress comes from repeatedly witnessing human suffering without being able to intervene directly. Over time, these demands can erode workers’ ability to cope and contribute to burnout and disengagement, especially where structured psychological support and recovery opportunities are not available[30].
Mental health risks from algorithmic management
Content review work is typically organised and monitored through algorithmic management systems that allocate tasks, track performance, and enforce productivity and accuracy targets in real time. Research on algorithmic management shows that such systems often intensify work pace, reduce worker autonomy, and increase surveillance, all of which raise psychosocial risks. Low remuneration per task drives work intensity and may even lead to competition amongst online content reviewers to access work. In content moderation, online content reviewers frequently report pressure to meet strict quotas for items reviewed per hour and to maintain high agreement rates with automated or benchmark decisions, with limited discretion or transparency about how metrics are calculated[31].
These forms of control and monitoring are associated with well-established psychosocial factors: work intensification, low decision latitude, and perceived unfairness, which in turn can cause stress, exhaustion, and reduced well‑being. Algorithmic task allocation can also lead to uneven distribution of particularly disturbing content, or to insufficient time for emotional processing between exposures, further aggravating mental‑health risks. When workers cannot understand how the system works or why decisions are made, they may experience feelings of powerlessness and unpredictability, which are recognised contributors to psychological strain[32]. Decisions often have to be made in a matter of seconds, sometimes thousands of times a day. The time pressure and the resulting workload increase stress[33]. The feeling of invisibility, the lack of organisational support, and the lack of appreciation and recognition exacerbate the negative effects[34].
Mental health risks from unfavourable employment conditions and structural vulnerabilities
Many online content reviewers work under employment conditions that amplify psychosocial risks: shift work, precarious contracts, limited control over schedules, and relatively low pay, especially in outsourced operations. Trade‑union and advocacy reports note that online content reviewers are exposed to daily levels of psychological harm that would be unacceptable in any other industry. Meanwhile, they face job insecurity and constrained opportunities to organise collectively. In business‑process outsourcing environments, long working hours and hyper connectivity, high workload, and restricted breaks are common, and online content reviewers may have limited access to employer‑funded mental‑health services or clinical supervision[35].
These working conditions also come with other recognised psychosocial factors such as high quantitative demands, irregular and unsocial working hours, low organisational justice, and insufficient social support at work. When combined with exposure to traumatic material, these factors increase the likelihood of chronic stress, burnout, and attrition; poor mental health is frequently cited as a major reason why online content reviewers leave their positions. Moreover, stigma around mental‑health and confidentiality requirements can create barriers to seeking help, leaving them to cope on their own in ways that may be insufficient or maladaptive[36].
Digital labour platforms may resort to microwork, whereby tasks are split up and distributed globally to a large number of individuals who can complete them asynchronously, providing platforms and their clients access to a large, flexible workforce capable of completing small tasks remotely[37]. Online content reviewing is associated with high levels of PSRs, and these risks are particularly acute when moderation tasks are organised as microwork via digital labour platforms (‘crowd work’ and ‘click work’). Content reviewers, including those working as platform microworkers, are repeatedly exposed to graphic violence, sexual abuse material, hate speech, and self‑harm content. Clinical and survey studies in commercial moderation show that 25-50% of online content reviewers score in ranges indicative of clinical depression or moderate‑to‑severe psychological distress, with clear dose–response relationships between frequency of exposure and levels of distress and secondary trauma symptoms. These findings are consistent with broader evidence on microworkers, among whom rates of depression can be 1.5-3 times those of the general population and about half report social anxiety[38].
Compared with in‑house or formally employed moderation, microwork‑based online content review tends to concentrate several high‑risk features: extreme content exposure, fragmented and intensively managed tasks, algorithmic control, and lack of OSH coverage. In the language of the Job Demands-Resources model, job demands (emotional load, time pressure, insecurity) are high, while resources (stable income, organisational support, mental‑health services, participation in decision‑making) are minimal or absent. This imbalance explains why online content reviewers and related microwork are increasingly highlighted in European and international reports as a critical frontier for regulating psychosocial risks in digital labour[39].
Prevention and protection measures
Policy and research point in the same direction: online content reviewing should be recognised as high‑risk work, OSH protections and preventive measures should be extended to self‑employed platform workers, and platform governance (including algorithms) should explicitly take account of PSRs and mental‑health impacts[40].
Here are concrete examples of prevention and protection measures for the mental health of online content reviewers, grouped by level of approach: organisation, management, individual:
General OSH and PSR policy frameworks
Because online content reviewers often work as self-employed contractors or platform freelancers, EU occupational safety and health (OSH) legislation generally does not apply to them. Hence, they are responsible for their own health and safety even if the client company has introduced safety measures[41]. Directive (EU) 2024/2831[42] on improving working conditions in platform work represents a significant step forward. According to Recital 40, “digital labour platforms should not, by means of automated monitoring systems or any automated system used to take or support decisions affecting persons performing platform work, process any personal data on the emotional or psychological state of persons performing platform work, (…)”. In addition, Article 12(1) states that “without prejudice to Directive 89/391/EEC and related directives in the field of safety and health at work, with regard to platform workers, digital labour platforms shall: (a) evaluate the risks of automated monitoring systems and automated decision-making systems to their safety and health, in particular as regards possible risks of work-related accidents, psychosocial and ergonomic risks; (…).”
The EU OSH and policy framework covers only employed workers, i.e. not the self-employed. The current legal framework does not ensure that self-employed online content reviewers can obtain legal recognition of the psychosocial harm caused by their work as work-related harm for labour and social security purposes. The general legal framework for psychosocial risk prevention is based on the Framework OSH Directive 89/391/EEC[43], whose Article 5(1) gives employers “a duty to ensure the safety and health of workers in every aspect related to the work”, which includes those to mental health even if not explicitly mentioned.
For both the self-employed and workers, it is reasonable to assume that Directive (EU) 2024/1760 on corporate sustainability due diligence applies when large companies are involved. This Directive is founded on EU’s values of respect for human dignity, freedom, democracy, equality, the rule of law and respect for human rights as enshrined in the Charter of Fundamental Rights of the European Union, and should include “fostering the sustainable economic, social and environmental development of developing countries”. According to Article 12(1) and (2), Member States shall ensure that companies take appropriate measures to identify and assess actual and potential adverse impacts arising from their activity. Taking into account relevant risk factors, companies must take appropriate measures to map their operations, those of their subsidiaries and, where related to their chains of activities, those of their business partners, to identify general areas where adverse impacts are most likely to occur and to be most severe and carry out in-depth assessment on this basis.
Prevention and mitigation frameworks to protect online content reviewers
UNI Global Union, the global union federation for the skills and services sectors, has defined eight protocols for sustainable content moderation[44]: which include exposure limits, realistic productivity quotas, mandatory trauma training, accessible counselling, stable employment and living wages, joint occupational health and safety and committees, enhanced migrant worker protections and the right to organise and collective bargaining.
The measures listed below have the potential to contribute to proper prevention of PSRs for online content reviewers. They imply a strong commitment from the companies and platforms using the reviewed content. They should also ensure, via technical and organisational measures, that workers do not bypass safeguards established with a view to earning more or accelerating their tasks. An adaptation of these measures to the type of relation between the reviewers and the user company (e.g. company setting with employees vs. microtask self-employed workers in a different country) is also necessary.
Work design and organisational measures
- Implement standardised psychological risk assessment and monitoring using validated mental health scales for exposed workers, with clear protocols for follow‑up and accommodation.
- Limit exposure timeto toxic content (rotation of tasks between neutral content and sensitive content, quotas/caps on hours per week/day/hour on the (most) burdensome content to review, alternation with non‑traumatic tasks).
- Build in frequent, mandatory breaks away from screens, with dedicated quiet or recreation spaces.
- Allow some flexibility in scheduling (flexible hours, partial remote work, possibility to move temporarily to lower‑risk queues) to accommodate individual coping capacities.
- Use technical protections: image/video blurring filters, automatic muting of audio, smaller preview windows, adjustable playback speed to reduce sensory impact.
- Involve reviewers in designing moderation rules and safety protocols, so they have more control and clarity, which reduces perceived helplessness and moral stress, including via joint occupational health and safety committees.
- Recognise content reviewing as a high‑risk job in internal policies and in employment contracts or collective agreements (additional risk indemnity, specific leave).
Leadership and management
- Train managers and team leaders to recognise signs of distress (sleep problems, avoidance, irritability, absenteeism, emotional numbness) and to respond quickly and non‑punitively.
- Set up regular psychological debriefings (group and individual) where reviewers can talk about difficult cases in a structured, confidential setting.
- Establish clear escalation protocols for particularly disturbing content, allowing reviewers to reassign or skip certain items without penalty.
- As appropriate, consider the availability of additional rest days or “recovery leave” after prolonged exposure to traumatic material or after critical incidents, as part of a voluntary code of practice or collective agreement.
- Provide workers with transparency on platform algorithms and involve workers in the design of algorithms.
Support, training and education on psychosocial risks and mental health
- Provide initial and continuous training to moderators and supervisors about the psychological risks of exposure to violent or abusive content (trauma, secondary traumatic stress, burnout) and about normal vs. alarming reactions.
- Train reviewers in evidence‑based coping strategies: cognitive‑behavioural tools to challenge intrusive thoughts, mindfulness, breathing techniques, grounding exercises.
- Inform staff about healthy digital boundaries and “hygiene”: limiting personal use of platforms they moderate, avoiding doom‑scrolling, etc.
- Provide 24/7 access to professional mental health support (on‑site or telehealth psychologists, workers assistance programmes, crisis lines). Guarantee confidential access with explicit separation from performance evaluation and HR disciplinary procedures to reduce fear of stigma or retaliation and ensure follow-up after the contract ends.
EU‑OSHA’s research focus on digitalisation has looked specifically into platform work and provides a range of publications indicating good practice for platform work[45]. Recent policy briefs emphasise the need to adapt PSR prevention to modern workplaces transformed by digitalisation, strengthen inspection and guidance, and combat stigma around mental‑health problems at work[46].
Conclusion
Online content reviewing has become an industrial-scale, largely outsourced and often precarious activity, frequently organised as microwork via digital labour platforms in low-wage regions, with no dedicated labour statistics to measure its extent. Reviewers face serious psychological risks, primarily due to repeated exposure to traumatic content (child sexual abuse material, graphic violence, hate speech, self-harm, etc.), work-related stress, and a lack of support but also due to being engaged in platform work, which comes with an additional range of risks. Addressing these risks requires more than compliance with general legal principles: it calls for concrete organisational measures and appropriate psychological support. In addition, policy could combine recognition of the activity as high-risk work, extension of OSH protections to platform and self-employed workers, stronger platform transparency, and organisational, managerial and collective-bargaining measures.
