You read through your feed. One post from a friend. There was an advertisement for shoes you were talking about once. A video that angers you -then another of that kind. One of your comments was deleted without a reason. Have you ever wondered: who made this decision? We are more likely to believe social media as an unbiased channel, merely relaying what our friends share. Nevertheless, that is not how it works. Algorithms are making decisions all the time, such as what to display, what to hide, what to mark as harmful, and what to boost. These unspoken rules control our digital life. This post states that algorithms are not neutral tools but actively control users by dictating our speech, our privacy, and even our safety, often without our awareness or consent. Hate speech and privacy violations are not distinct issues, but they are effects of the nature of the algorithm system.
The Myth of Neutrality: Algorithms as Rule-Makers
The majority of individuals envision algorithms as cold, logical, fair, and simple math. Sum the likes, rank by popularity, complete. But that’s a myth. Media critic Mark Andrejevic (Andrejevic & Volcic 2019) refers to this as automated culture, the conception that our feeds are not choosing what is out there to show us, but rather pre-determining what we see. Imagine a librarian who does not allow you to window shop. Instead, they race away and put some books in your hands, according to their ideas of what you will desire. Now, suppose that the librarian gets to determine also what books do not even get on the shelf. This is what YouTube and TikTok algorithms do in a second. They do not merely arrange the content, they do produce your reality, piece of advice at a time. Artificial intelligence is never merely technical, as AI researcher Kate Crawford (Crawford, 2021, Atlas of AI, pp 1-21) asserts. It is associated with actual political expenses: who is perceived, who is muted, and who is spied upon. It is not a glitch when the algorithm of YouTube leads you to more and more radical videos, trying to keep you watching. That’s a design choice that benefits engagement metrics – not your wellbeing. This demonstrates that algorithms are in effect controlling what we view before we can even realize that we had a choice.
But now comes the part where it is still more misleading. Algorithms not only feed you with something, but they also possess power, concealing what you do not see. In his platform governance, sociologist Tarleton Gillespie notes that algorithms are custodians in that they are the ones that silently determine which posts pass the threshold of visibility and which vanish into the depths. You may believe that you are scrolling through everything possible. You’re not. Thousands of posts, videos, and comments have already been filtered by the algorithm following the patterns it has learned about your previous behaviour. Unless you’ve ever clicked on a news article about climate change, the algorithm no longer shows you such an article, not because it is irrelevant, but because the algorithm has classified you as having a low engagement risk with that subject. With time, your feed becomes small. You watch what you already like and less of what challenges you. This is not censorship as we know it. It is an algorithmic rule of invisibility. The algorithm never requires silencing a video by banning it. It only has to hide it deep in your feed so that you never scroll up far enough to see it. And since you never know what you are missing, you never consider asking. Algorithms rule like that, silently, continuously, and invisibly.
The only real power of this is that the majority of users are unaware of it happening. We trust the feed. We presume that it is what it looks like. However, as Crawford cautions, the political price of this faith is very high. Algorithms favour extreme content when they are the ones that are oriented towards engagement rather than accuracy. Outrage prevails when they make watch time more important than diversity. When they place retention above safety, harm prevails – silently, in the background, optimised by lines of code that are not subject to a vote and cannot be observed. The algorithm is not an unbiased referee. It is a rule-maker that has got its own ax to grind: scroll away, at any rate. And until we realise that we are not really making the choice we perceive. We are getting what the algorithm has already selected for us.
How Algorithms Reshape Privacy – Beyond Just “Consent”
Is it a violation when WeChat scans your personal messages to be monitored on security grounds? The majority of Western users would respond yes. It could be accepted by many Chinese users. This distinction is not due to the fact that one group is concerned with privacy and the other is not. The reason is that algorithms determine the concept of privacy in the real world. Philosopher Helen Nissenbaum (Nissenbaum, 2018) refers to this as contextual integrity; privacy is not about concealing anything, but about information flowing in the right way in the right context. But context is not automatically respected by the algorithms of platforms. They are programmed to gather, analyse, and distribute data depending on corporate or state interests. According to a study by Chen & Cheung (2018), the participants of WeChat are willing to sacrifice privacy in an attempt to achieve convenience since the logic of the algorithm encourages engagement – the more information one provides, the more convenient the service will become. Legal theorist Nicolas Suzor refers to such governance as lawless: platforms create internal rules secretly, with no input from the democratic process (Suzor, 2019, pp 24). This is an illustration of how algorithms actively construct the definition and experience of privacy not as a right, but as an attribute to optimise or trade off.
How Algorithms Amplify Hate Speech and Propaganda – Engagement Over Safety
You would imagine that propaganda and hate speech are two totally separate issues. The problem is, however, that they operate under the same algorithmic engine. Researchers Bolsover & Howard (2019) studied Chinese computational propaganda on Twitter and Weibo. They found that automated bots amplified political messages on both Chinese and Western platforms. Ideology is irrelevant to the algorithm. It merely takes orders to reach the maximum. This is what happened to Reddit during Gamergate, as researcher Adrienne Massanari recorded (Massanari, 2017, 19(3)). The upvote/downvote algorithm used by Reddit to promote the best content on the platform to users actually increased the toxic harassment. Mobs posted threats to the front page and voted. The algorithm was unaware of the fact that the protest hashtag and the death threat were different. It has just been engaged: clicks, votes, comments. The more the outrage, the more visibility. This shows how algorithms are giving precedence to engagement rather than user safety. In the meantime, Sinpeng et al. (2021) discovered that Facebook primes its automated moderation with uneven rules in countries. The algorithm does not get local context or slurs. It simply coincides with trends – and trends are simple to cheat on by both propagandists and harassers. The algorithm doesn’t know the difference between a protest hashtag and a harassment campaign. It merely observes involvement.
The more significant issue that connects the concept of propaganda and hate speech here is that algorithms are indifferent to harm structurally. The dissemination of election misinformation by a bot network and the coordinated campaign of harassment with death threats is equally generating valuable currency: engagement. Likes, comments, shares, watch time. The algorithm doesn’t feel outrage. It doesn’t recognise trauma. It merely quantifies activity and punishes that which results in an increase in activity. In her theory of surveillance capitalism, Zuboff (2019) proposes that platforms harvest human experience as free raw material, our clicks, our scrolls, our outrage, and process it through prediction algorithms, which are sold access to our future behaviour. When a hateful post goes viral, the algorithm is not failing. It is doing just as it was intended: it is pushing whatever makes the users fixate on their screens. Bugs in the system are not propaganda and hate speech. They are byproducts of a logic of algorithms that considers all engagement good engagement. The algorithm will keep rewarding the worst of us, not the best, until we redesign that logic, making safety, accuracy, and well-being its priorities, rather than retention.
It’s further complicated because the very attitude of indifference is inherent to the model of these companies. Fact-checking people are expensive. Moderators suffer from burnout. Algorithms that spread outrage, on the other hand, generate infinite free energy for the machine of engagement.
A Current Case Study: TikTok, Algorithmic Governance (2025-2026).
We can take this home on one of your likely daily platforms: TikTok. Late in 2025, internal leaks revealed that the TikTok algorithm that was used to promote content to users on the For You Page was actually specifically targeting more extreme content to those who spent over 90 minutes on the platform. The logic? Extreme content will lead to increased engagement – more watch time, more shares, more returns. But the results were dire. Adolescents said they were being pressured into eating disorder content, political radicalisation, and hazardous pranks. After researchers attempted to replicate the effect, they discovered that the algorithm indeed actively guided users to more extreme versions of whatever they originally watched. Watch a single fitness video? You’ll see crash diets. Watch a political clip? You’ll see conspiracy theories. This example emphasizes the fact that platform governance is inherent in algorithms, not as a bug, but as a feature that is necessary to maximize attention by any means. In early 2026, the European Union initiated an inquiry into whether the algorithm of TikTok contravenes the need to alleviate systemic risks in the Digital Services Act. TikTok’s response? It introduced additional protection measures, which proved to be additional filters in its algorithms that could be easily circumvented by users. The algorithm was not predetermined. It was remodeled to appear fixed and maintain engagement. This is algorithmic governance at work: non-humane, mechanistic, and extremely hard to refute. It is the same with TikTok, YouTube, and X (previously Twitter). Algorithms not only mirror what we desire, but also act to influence what we desire by filtering what we watch.
Conclusion: Who’s Really Holding the Keys?
Platforms are not merely reflections of society, but algorithmically based on what we are allowed to say, what we make private, and who is hurt. We have been shown how algorithms pre-determine our reality (Andrejevic & Volcic, 2019), how they revise the concept of privacy beyond mere consent (Nissenbaum; Chen and Cheung), how they promote hate speech and propaganda by structuring to govern (Bolsover & Howard, 2019; Massanari, 2017; Sinpeng et al., 2021), and how the newest design decisions of TikTok are determined to govern the code. In all of the cases, there is one trend: algorithms are not neutral. Each design decision, such as upvote buttons or the For You Page recommendations, is a manifestation of someone prioritizing. Typically, they happen to be a corporation that is optimising to obtain your attention, but not your health. There, then, is your lesson. The next time an application requests permission, a post is flagged, or a video goes viral due to the wrong reasons, stop and ask two questions: who created this rule, and to whom is it beneficial? The answer won’t always be clear. That’s the point. These algorithmic mechanisms are aimed to be transparent. But you don’t need to be a coder or a policy expert to stay curious. All you need to do is to keep in mind: your feed is not a window. It’s a machine. And there is another one with the keys. The only real question left is: When will we start asking for them back?
References
Andrejevic, M., & Volcic, Z. (2019). From mass to automated media: Revisiting the ‘filter bubble’. In N. Persily, M. Tucker, & J. Richman (Eds.), Big data, political campaigning and the law. Routledge. https://www.taylorfrancis.com/chapters/edit/10.4324/9780429288654-2/mass-automated-media-mark-andrejevic-zala-volcic
Bolsover, G., & Howard, P. (2019). Chinese computational propaganda: automation, algorithms and the manipulation of information about Chinese politics on Twitter and Weibo. Information, Communication & Society, 22(14), 2063–2080. https://doi.org/10.1080/1369118X.2018.1476576
Chen, Z. T., & Cheung, M. (2018). Privacy perception and protection on Chinese social media: a case study of WeChat. Ethics and Information Technology, 20(4), 279–289. https://link.springer.com/article/10.1007/s10676-018-9480-6
Crawford, K. (2021). Atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press. (pp. 1-21). https://doi.org/10.1162/leon_r_02206
Gillespie, T. (2018). Custodians of the internet: Platforms, content moderation, and the hidden decisions that shape social media. Yale University Press. https://doi.org/10.12987/9780300235029
Massanari, A. (2017). Gamergate and The Fappening: How Reddit’s algorithm, governance, and culture support toxic technocultures. New Media & Society, 19(3), 329–346. https://doi.org/10.1177/1461444815608807
Nissenbaum, H. (2018). Respecting context to protect privacy: Why meaning matters. Science and Engineering Ethics, 24(3), 831852. https://link.springer.com/article/10.1007/s11948-015-9674-9
Sinpeng, A., Martin, F. R., Gelber, K., & Shields, K. (2021). Facebook: Regulating hate speech in the Asia Pacific. Department of Media and Communications, The University of Sydney. https://appap.group.uq.edu.au/files/1779/2021_Facebook_hate_speech_Asia_report.pdf
Suzor, N. P. (2019). Lawless: The secret rules that govern our digital lives. Cambridge University Press. (pp. 10-24) . https://doi.org/10.1017/9781108666428
Zuboff, S. (2019). The age of surveillance capitalism: The fight for a human future at the new frontier of power. PublicAffairs. https://www.proquest.com/openview/05487de4e801c1e46aa0480bde8248ad/1?pq-origsite=gscholar&cbl=2035668
Be the first to comment