When AI decides what you can say online

In the digital world, where everyone can have a voice, are comment filtering algorithms helping to protect the community, or are they inadvertently stifling the voices of users themselves?

AI is scanning and deleting comments on social media.

When users leave a comment under a YouTube video , a Facebook post, or an online article, they're not sure if that comment will reach the intended audience. This isn't because it's misspelled or illegal; it's simply because it's blocked by an automated filtering algorithm.

Current technology platforms employ comment filtering systems to remove hate speech, pornography, violence, and divisive language. The goal is to "keep the community safe," but could these tools be cleaning comments to the point of becoming insensitive?

When AI becomes the gatekeeper of speech

Unlike the early days of the internet, where "raw" comments were freely available, most major platforms now delegate content filtering to artificial intelligence. Thanks to machine learning, these systems can scan millions of lines of text daily, evaluating language, tone, and even context to decide whether to keep or delete content.

However, this system is not always accurate.

Ms. Ngoc Vy (Ho Chi Minh City) shared: "My comment and feedback on a product was hidden for no apparent reason. There was no cursing, no spam, just a comment, but it disappeared after a few seconds."

Many pieces of content are removed simply because they use sensitive words in a perfectly normal context. Phrases like "poverty," "hunger," "policy," or "system" are sometimes misinterpreted as negative if the machine doesn't understand the meaning of the sentence.

Filter out or suppress dissent?

According to a warning from the German organization AlgorithmWatch, automated censorship systems can inadvertently silence dissenting voices, especially from marginalized or minority groups, because these systems learn from the behavior of the majority, which lacks sufficient cultural and ideological diversity.

In Vietnam, the strict control of content by social media platforms is becoming increasingly common, especially on highly interactive news sites. Comments on current events and social issues are often restricted or require "pre-approval," leading many to question: Does freedom of speech still exist online?

The worrying aspect is that users rarely know why comments are hidden or deleted. There are no notifications, no transparent feedback mechanisms. Everything happens silently, and in an algorithmic world , silence can be the most subtle form of censorship.

Keep the internet clean, but don't whitewash the dialogue.

A clean online space should not be synonymous with a "sterile" space. Society is a collection of differing opinions, constructive criticism, and debate. We need algorithms to filter out harmful comments, but we also need to protect the right to speak differently, the right to question, and the right to raise concerns.

When every comment has to go through a "screening process," we must ask ourselves: Is technology protecting the community, or is it selectively choosing what the community is allowed to hear?

Technology cannot replace humans in making value-based decisions. And in the age of AI, transparency and the right to feedback are the fairest "algorithms" for the entire system.

Technology should help people voice their opinions, not decide for them what to say.

SONG TRI

Source: https://tuoitre.vn/khi-ai-quyet-dinh-ban-duoc-noi-gi-tren-mang-20250701231035288.htm