Social media algorithms know about your sexual orientation before you do. This is called "algorithmic outing" – and it can be dangerous

In the era of social media dominance, our perception of privacy has changed dramatically. For the LGBTQ+ community, the internet has long been a "safe haven," a place to explore identity and find like-minded people. However, today the digital space is turning into a mirror that not only reflects us, but also reveals our deepest secrets before we are even ready to share them.

Today, social networks are increasingly determining the sexual orientation or gender identity of users even before they realize it themselves or decide to come out. In a new study published in the journal Gender, Place & Culture, a special term was proposed for this phenomenon — "algorithmic outing".

We tell you how it works and why the invisible operation of algorithms in public places can become a threat to security.

How does the algorithm understand who I am?

Modern platforms like TikTok no longer rely on your likes alone. Their algorithms analyze "engagement signals": which videos you linger on, which authors you watch to the end, and even how often you return to a certain type of content.

The lead author of the study, Justin Ellis, coined the term "algorithmic outing" to describe a situation where a digital system identifies and classifies a user without their knowledge or consent. One of the study participants, a 25-year-old lesbian, admitted that the algorithm began offering her queer content when she herself was still in the process of self-discovery and used an anonymous profile.

Artificial Intelligence and "Digital Physiognomy"

Concerns about privacy have intensified after the publication of a high-profile study by scientists from Stanford University. They trained a neural network to determine sexual orientation from photos of faces taken from dating sites. The results were frightening: алгоритм правильно определил ориентацию в 91% случаев для мужчин и в 83% для женщин, significantly exceeding the capabilities of a person.

Although critics called it "AI phrenology" and pointed to the limitations of the sample (only white people were used), the very possibility of such an analysis poses a huge threat. In the hands of repressive regimes where homosexuality is criminalized, such technologies can be integrated into street video surveillance systems to automatically identify "gender-atypical" individuals.

Shadow Profiles: You're Not Alone in Your Loss of Privacy

Many people believe that by deleting their account or not providing personal information, they are protected. This is a misconception. There is a phenomenon of "shadow profiles" (shadow profiles) — databases that social networks create on people who don't even have accounts there.

Your privacy is a collective phenomenon. If your friends share their contact lists or photos where you're tagged, algorithms can calculate your orientation with a high degree of probability based on connections within your social group.

Even a complete absence from social networks does not guarantee privacy if your loved ones actively use them.

What is the main danger?

Today our life takes place in the so-called hybrid spaces, where online activity intersects with physical reality. This creates specific risks:

• Screen publicity: When you scroll through the feed on the bus, the algorithm can display specific queer content on the screen at the exact moment it is seen by a random passerby.
• Lack of context: The systems do not take into account your personal readiness to come out or the level of security in a particular place (for example, in a conservative family or at work).
• Digital Surveillance: apps like Grindr have been repeatedly criticized for sharing user data (including HIV status and geolocation) with third parties without explicit consent.

How to protect yourself in the digital world?

In the face of growing hostility and the use of algorithms to sort people by type, the LGBTQ+ community is forced to develop new coping strategies. Here are a few tactics that users use today:

1. Passive consumption: Viewing content without likes, comments, and follows so as not to give the algorithm training data.
2. Multiple accounts: Using different profiles for different purposes (personal, activist, research).
3. Anonymity and avatars: Using profiles without real names and photos during the identity research phase.
4. "Quick escape": A request to implement features to instantly hide content on the screen if your personal space is suddenly invaded by strangers.

Tech companies must be held accountable for the way their algorithms reveal sensitive information. In the meantime, our "digital mirror" remains not only a tool of cognition, but also a potential weapon against us.

The researchers are calling on tech companies to be more transparent and accountable. Among the proposed solutions is the implementation of the principles of "privacy by default" (privacy-by-design) and the creation of tools that will allow the user to quickly hide sensitive content if they suddenly find themselves in an unsafe situation.