Misinformation and disinformation are not the same
The two words are often used interchangeably, but the distinction matters for how you moderate. Misinformation is false or misleading content shared without the intent to deceive, a reader passing on something they believe is true. Disinformation is false content created and spread deliberately to manipulate. The first is a mistake. The second is a strategy.
A newsroom comment and debate space sees both. Treating them identically, by deleting everything that looks wrong, is how you lose the legitimate debate along with the bad-faith content. The goal is not a silent comment section. It is a space where good-faith disagreement thrives and coordinated manipulation does not take hold.
Why a debate space is exposed, and why it matters
An open conversation under a news article is, by design, a place where claims circulate fast. That is its value, and its risk. Around elections, health stories or polarising policy moments, a contributory space can become a target for coordinated false claims, brigading and bad-faith framing.
For a publisher, the stakes are direct. Unchecked disinformation erodes the trust that the masthead spent decades building, drives away the loyal readers you most want to keep, and now carries regulatory exposure under the Digital Services Act. Moderation is no longer a cost center. It is part of protecting the brand.
The wrong fixes: shutting down, or generic filters
Two common responses both fail. The first is closing comments entirely. That removes the symptom and the value at once, handing the conversation about your journalism to social platforms you do not control.
The second is bolting on a generic toxicity filter built for social media. Those models, tuned for slurs and harassment, underperform on press content. They flag opinion-heavy, argumentative writing as toxic and miss the calm, well-written false claim that actually does the damage. Press debate needs moderation trained on press debate.
How moderation holds the line: AI plus human, at scale
The workable model is hybrid. AI moderation handles the volume, and human editors handle the judgment calls.
In practice, Logora's AI filters around 85 percent of toxic content before it ever reaches the human queue. The model was trained on roughly 45,000 human-labeled examples and has moderated more than 50 million reader contributions since 2019, across French, English, German, Spanish, Portuguese and beyond. Your team reviews the remaining share, where context and intent matter most.
The effect on quality is measurable. At Milenio, the share of approved contributions rose from about 60 percent before Logora to 80 to 85 percent after. A better-moderated space is not a quieter one. It is one where more good contributions get through and fewer bad ones survive, which is exactly what keeps a structured debate civil enough to be worth joining.
The DSA angle: moderation you can defend
Under the Digital Services Act, moderation is not just an editorial choice, it is a documented obligation. Every removal needs a statement of reasons (Article 17), and platforms owe transparency reporting (Article 24). Moderation done in spreadsheets cannot meet that bar.
The defensible approach is to journal every decision at the moment of submission, deliver the statement of reasons in the user's own language, and be able to export an annual transparency report in minutes rather than weeks. Pair that with a clear, public moderation charter so readers know the rules before they post. Compliance and a healthy debate are the same project.
Conclusion: defend the debate, not just the thread
Disinformation is not a reason to close the conversation. It is a reason to run it well. The combination of press-trained AI moderation at scale, human judgment on the hard cases, and DSA-grade documentation lets a newsroom keep its debate space open, civil and trustworthy, without drowning its team or its credibility.
To see how the moderation pipeline works in practice, start with the AI moderation module or the comment software overview.