Goodbye humans, hello algorithms.” This phrase could summarize TikTok’s latest and controversial decision: replacing hundreds of human moderators with artificial intelligence systems capable of detecting and eliminating harmful content in a matter of seconds.
What seems like a logical step towards technological efficiency is actually a movement that opens a debate about the future of internet moderation. Is AI reliable enough to decide what we see and what we don’t see on social media?
The case of TikTok is not isolated: automation is advancing on all digital platforms, driven by new regulations and the search for lower costs. However, are we safer with machines at the helm or more vulnerable than ever?
TikTok makes the leap to automation
TikTok announced the dismissal of around 300 employees from its London office of the trust and safety department. Their work, focused on reviewing videos and deciding whether or not they met the platform’s rules, will be replaced by AI algorithms.
It is not an isolated event, but rather part of a global reorganization that has already affected other offices in Berlin and Malaysia, where hundreds of positions were also cut. The company argues that this decision is supported by advances in its automated systems.
According to official data, TikTok’s AI removes 80% of content that violates guidelines and detects 99% of dangerous material even before a user reports it.
In addition, it ensures that automation reduces the exposure of the remaining moderators to violent or traumatic scenes, presenting it as a labor protection measure.
The timing is not coincidental: the entry into force of the UK Online Safety Act, which requires the rapid removal of illegal content and provides for million-dollar fines, has accelerated the company’s technological commitment.
The numbers behind the decision
In 2024, the platform’s revenue in Europe grew by 38%, reaching $6.3 billion, while losses fell from 1.4 billion in 2023 to just $485 million in the last year.
Far from being in crisis, the company is going through a moment of progress that allows it to invest in new tools. In fact, it has announced a $2 billion plan in 2024 aimed at strengthening its AI-based moderation systems.
The official narrative is clear: more efficiency, less human risk and faster compliance with demanding European regulations. But unions and analysts offer another reading.
According to the Communication Workers Union (CWU), behind the discourse of innovation hides a strategy of cost reduction and relocation.
While a moderator in London could earn around £2,000 a month, subcontracted workers in countries like Colombia receive salaries eight times lower for similar tasks.
The role of artificial intelligence in content moderation
Platforms like TikTok, Facebook or YouTube use trained algorithms to analyze millions of posts in real time and detect signs of violence, nudity, hate speech or misinformation.
Its main advantage is scale: no human team could review such a volume of content with the same speed. The AI also learns from previous patterns and improves with each interaction.
This allows you to identify suspicious behavior and take action even before a user reports a video. In theory, this anticipation reduces the spread of harmful content and protects the most vulnerable audiences.
However, it’s not all lights. Artificial intelligence has difficulty understanding the cultural and linguistic context of certain messages. A word considered offensive in one country may be innocuous in another, and an ironic meme may be hate speech.
These errors generate both false positives (content unfairly removed) and false negatives (harmful posts that go unnoticed).
A future without human moderators?
TikTok’s commitment to automation has ignited a key question: can digital platforms completely do without human moderators? In London, Berlin and Malaysia, cuts are already a reality.
However, there are still regional centers in Dublin and Lisbon that maintain human teams in charge of verifying complex cases, indicating that total replacement is not yet possible.
International examples show that this is not an isolated process. In the technology industry there is a clear trend towards outsourcing and automation of moderation, both to reduce costs and to comply with stricter regulations.
However, experience shows that there are areas in which AI remains insufficient: the detection of ambiguous political speeches, well-crafted fake news or messages that strongly depend on the social and cultural context.
Benefits of AI in moderation
The use of artificial intelligence in social media moderation has advantages that explain why companies like TikTok are betting so heavily on it:
- Unprecedented speed and scale: AI can review millions of posts in seconds, something impossible for a human team. This makes it possible to stop the spread of harmful content before it reaches high visibility.
- Greater efficiency in regulatory compliance: Faced with laws such as the UK Online Safety Act, which requires removing illegal material quickly, algorithms offer platforms the ability to respond in real time and avoid million-dollar fines.
- Reducing exposure to traumatic material: Manually moderating violent or disturbing videos can affect workers’ mental health. With AI, much of this content is automatically filtered, reducing the psychological burden on human teams.
- 24/7 Availability: Unlike humans, automated systems never rest, which guarantees constant monitoring in all time zones and in different countries.
- Continuous learning capacity: Thanks to training with large volumes of data, the algorithms improve over time, refining their detection criteria and becoming more precise.
Advance or setback for social networks?
The presence of AI in content moderation marks a turning point in social media interaction. It is a decision that redefines the responsibility of platforms for what users consume and how they feel protected.
While AI offers speed, continuous availability, and less exposure to traumatic material for humans, it also raises ethical dilemmas about censorship, bias, and the loss of critical judgment that only a moderator can provide.
The real challenge is finding a balance between technological efficiency and human judgment, ensuring that networks remain safe, transparent and culturally sensitive spaces.
The path to ideal moderation is not to remove people, but to integrate algorithms and human oversight in a way that maximizes protection without sacrificing understanding and context.
This post is also available in: