Disinformation: why a correction is not enough

Correcting is not enough

A correction is necessary when false information circulates. It is almost never sufficient. The reason is simple: a message is not corrected the way a calculation error is. It addresses audiences that already have a history with the subject, beliefs, affiliations and sources of trust.

Work on disinformation communication has identified several traps. The most counter-intuitive is repetition: the more a myth is heard, the more familiar it becomes, and that familiarity can be taken for truth, even by people who do not endorse it. A correction that begins by setting out the false information at length therefore works against itself.

Why corrections fail

Three mechanisms explain the most frequent failure.

Familiarity. Repeating a claim, even while denying it, makes it easier to recognise. That fluency is often read as a cue of truth. The myth stays in memory, the correction fades.

Identity. A correction that contradicts a belief tied to a group's identity is often received as an attack. The audience does not reassess the fact: it reassesses the source.

The messenger. An organisation perceived as a party to the matter has less credit than an independent source. The same fact, carried by a recognised third party, can be accepted where the interested party's word is rejected.

These mechanisms do not condemn correction. They require it to be built differently.

The truth sandwich

The Debunking Handbook 2020, a synthesis of research on correcting false information, recommends a simple structure, often called the "truth sandwich".

The fact first. The first sentence gives the accurate information, in a way that stands on its own and is understandable out of context.

The myth next, once only. It must be named so that the audience knows what is being discussed, without repeating it in every paragraph.

The fact again. The conclusion returns to the accurate information, so that the last thing retained is the right one.

The technique explained. Why the message is false is explained: a truncated figure, an invented source, a correlation presented as a cause, an image taken out of context. This step is decisive: it does not only correct a content, it teaches a process to be recognised.

This structure is not a communication recipe. It is a way of protecting the correction against the repetition effect.

Preventing rather than correcting: inoculation

Another route consists in acting before the spread. Psychological inoculation, or prebunking, exposes the public to manipulation techniques in a controlled way, so that it recognises them later.

A study published in Science Advances tested five short videos on frequent techniques: emotional language, incoherence, false dilemma, scapegoating and ad hominem attack (Roozenbeek et al., 2022). The authors ran seven preregistered studies, including six randomised trials with 6,464 participants and a field study on YouTube with 22,632 people. The videos improved recognition of the techniques, confidence in the ability to spot them, and the quality of sharing decisions.

Two lessons for an organisation. First, prevention works better than late correction. Second, it addresses techniques, not only contents: that is what makes it transferable from one subject to another.

What European law changes

The Digital Services Act, or DSA, has applied to all intermediary services since 17 February 2024. It imposes transparency obligations, systemic risk assessments and mitigation measures on platforms, particularly very large ones.

In France, Arcom is designated as digital services coordinator. VIGINUM, a service of the General Secretariat for Defence and National Security created in 2021, monitors foreign digital interference and publishes analyses of information campaigns.

This framework does not relieve an organisation of building its own response. It does provide counterparts, procedures and levers when disinformation involves a platform or foreign interference.

What an organisation can do

Five steps, in this order.

Measure. Before answering, know what is circulating, where, with whom and how fast. A marginal rumour does not deserve a public response.

Decide. Answer, do not answer, answer through a third party, file a complaint, refer to a platform: the options are not equivalent and do not exclude one another.

Correct with the right structure. The fact first, the myth once, the fact again, the technique explained.

Equip those who speak. Teams, partners and spokespeople need talking points built on the same mould, not contradictory messages.

Measure the effect. A correction is not judged by the number of views, but by how the narrative evolves: who takes up what, with which words, in which space.

What it changes in a crisis

Disinformation is not a communication accident. It is a balance of power, played out on credibility, speed and trust. An organisation that simply publishes a press release leaves others to define the narrative.

A useful response combines three things: a factual and fast word, an explanation of the processes, and credible relays. The rest is a question of resources and timing.

On this ground, see crisis communications. For the handling of the first hours, see The first three hours of a crisis.