Homophobia Is Rising: How Social Media and AI Amplify Hate

Carle Jasmin (Image : Pixabay)

For the past few months, we have been witnessing a spectacular resurgence of homophobia, particularly on social media. Let us not be mistaken: I am not saying that this homophobia is not real, since it has found followers who physically and concretely target LGBTQ+ people in several cities across the United States, including two recent murders in gay clubs, as well as attacks in Canada and Europe.

But this new wave of homophobia seems to be driven primarily by a bandwagon effect, propagated by a handful of individuals who multiply hateful messages and calls for aggression, notably by using artificial intelligence to mass-produce and distribute content. An explanation of this theory.

For several years, and particularly since Donald Trump returned to the presidency of the United States in January 2025, hostile rhetoric toward LGBTQ+ people appears to have regained a troubling level of visibility in several regions of the world. But can we really speak of a global resurgence of homophobia? The available data allow us to answer yes on certain indicators, but with an important qualification: there is currently no global statistic that would allow us to directly attribute a measurable increase in homophobia to Donald Trump.

The most recent data nevertheless show something that is difficult to ignore: some indicators of acceptance of LGBTQ+ people are declining, while hostile political rhetoric and certain restrictive measures are multiplying.

One of the most interesting international indicators comes from the Ipsos LGBT+ Pride 2026 survey, conducted in 26 countries. According to this study, 66% of respondents say they support same-sex marriage or its legal recognition.

The figure remains a clear majority, but it represents a decline from the previous year. In 2025, the proportion was 69%. Over the course of one year, support therefore fell by 3 percentage points. This decline obviously does not mean that 3% of the world’s population suddenly became homophobic. The survey covers only 26 countries, and year-to-year variations can be influenced by many political, social and cultural factors.

Nevertheless, it is an interesting statistical signal: after decades of generally increasing acceptance of same-sex couples in several Western societies, some indicators are beginning to show movement in the opposite direction.

What if a few individuals were enough to create the impression of an explosion of homophobia?

Social media have profoundly changed the way hatred spreads. Today, a handful of extremely active accounts can make a minority appear to be a majority. With automation and artificial intelligence, the phenomenon could reach an unprecedented scale.

For the past few months, we have been witnessing particularly high visibility of homophobic and anti-LGBTQ+ rhetoric on social media. Insults, accusations, calls for exclusion and sometimes even calls for violence circulate daily on the major platforms. But one question deserves to be asked: how many people are actually behind this massive volume of content?

Research into social media manipulation shows that a relatively small number of highly active accounts can produce a disproportionate amount of content and, above all, generate amplification far greater than their actual numbers. The mechanism is relatively simple.

Imagine ten particularly active people. If each of them posts 20 messages a day, that already amounts to 200 daily posts. If those messages are reposted, commented on or reformulated by just 20 other accounts, the number of posts associated with the same narrative quickly reaches several thousand. And that is only the beginning.

Each post can be seen by hundreds, thousands or even hundreds of thousands of people. The number of people exposed to the message then becomes completely disproportionate to the number of people who originally produced it. Ten people can therefore produce ten voices, but the Internet can give those ten voices an audience of millions of people. That is where the problem begins.

On the Internet, the apparent popularity of an idea does not necessarily correspond to the actual number of people who support it. A user who successively views several anti-LGBTQ+ messages can easily get the impression that “everyone thinks this way.” Yet these messages may come from a very limited number of accounts that cite one another, respond to one another and systematically repost the same content.

The phenomenon is sometimes described as a form of coordinated amplification or “astroturfing,” when activities originating from a small group artificially create the impression of a much broader movement. The mechanism is particularly effective when the same message is reformulated several times. One post can be transformed into a video, an image, a comment, a humorous post or a purported news story. The content then appears to be new even though it is essentially based on the same original narrative.

Social media platforms do not necessarily display posts in chronological order. Platforms seek, among other things, to determine which content is likely to hold their users’ attention. Content that triggers a strong emotional reaction often has greater engagement potential. Anger, outrage, fear or disgust can therefore become extremely effective drivers of dissemination.

A particularly provocative homophobic message can thus generate hundreds of comments, including some from people condemning the message. But from a visibility standpoint, the result remains the same: the content receives more interactions and potentially becomes more visible. The paradox is striking: even people who are fighting a hateful message can sometimes inadvertently contribute to its spread when they repost it in order to denounce it.

There is another problem as well: not all accounts on social media necessarily correspond to a real person acting independently. Manipulation campaigns can use automated accounts, fake accounts or multiple accounts controlled by the same group. Their objective may be to artificially multiply interactions around content in order to give it greater credibility or visibility.

A university study on coordinated social media manipulation has shown how small groups can use multiple accounts to artificially amplify certain messages and create the impression that an opinion is much more widespread than it actually is. Technology therefore makes possible something that would have been extremely difficult to accomplish just a few years ago: simulating the appearance of a popular movement.

And this is where artificial intelligence could profoundly change the situation.

Before the arrival of generative AI tools, someone wishing to disseminate the same message in different forms had to write each post themselves. Today, artificial intelligence tools can quickly produce numerous variations of the same content: different texts, different headlines, translations, summaries, responses to comments or versions adapted to different platforms.

AI can also facilitate the production of images and videos that present the same narrative in different forms. Artificial intelligence does not necessarily create hatred. Above all, it can dramatically reduce the cost and time required to produce and disseminate it.

A particularly revealing example in Canada

The phenomenon is not merely theoretical. In March 2026, an analysis by the Institute for Strategic Dialogue examining anti-LGBTQ+ rhetoric in Canada documented an explosion in the volume of hateful content following the Tumbler Ridge shooting.

According to this analysis, more than 10,000 anti-LGBTQ+ posts were recorded in Canada in a single day. The organization also observed that the Canadian extremist accounts studied had increased from approximately 20 daily posts to 233, an increase of more than 1,000%.

The analysis showed that a few recurring narratives had generated a significant proportion of the engagement. In other words, a crisis can become the starting point for a genuine digital avalanche, in which a limited number of actors massively produce or relay certain narratives.

The phenomenon is all the more concerning because a post does not need to be endorsed by those who see it in order to be effective. One person may share content to condemn it. Another may post it in a group to ask for an explanation. A journalist may report on it. A politician may respond to it. An influencer may comment on it. At each stage, the content reaches new audiences. This is how a message initially posted by a handful of people can ultimately be seen by millions of users.

The reach of hateful rhetoric therefore does not necessarily measure the number of people who share it. It also measures the ability of the digital system to circulate it.

However, it would be dangerous to conclude that digital homophobia is merely an illusion. Attacks, threats and hate crimes do exist. LGBTQ+ people are genuinely targeted, and some digital campaigns can have very real consequences in the real world. The issue is rather to understand how a minority can use new technologies to amplify its ideas to the point of influencing the broader social climate.

The boundary between the Internet and reality no longer truly exists. A discourse repeated often enough can help normalize certain words, accusations and prejudices. What once seemed marginal can eventually enter mainstream political or media language.

In the face of homophobic hatred, whether it is directed personally at us or generalized, we must not remain passive. We must fight it as quickly as possible, before it becomes entrenched in the imagination of those who hate, who often seek nothing more than validation of their ideas and prejudices, eventually giving themselves a justification to target who we are and what we represent.

ADVERTISING

What if a few individuals were enough to create the impression of an explosion of homophobia?

A particularly revealing example in Canada

Leave a Reply Cancel reply