Algospeak

For the 2025 book, see Algospeak (book).

In social media, algospeak is the use of coded expressions to evade automated content moderation. It is used to discuss topics deemed sensitive to moderation algorithms while avoiding penalties such as shadow banning, downranking, or de-monetization of content. It is a type of internet slang^[1] and a form of linguistic self-censorship.^[2][3] The term algospeak is a portmanteau of Algorithm and -speak;^[4] it is also known as slang replacement or Voldemorting,^[3] referencing the fictional character also known as "He-Who-Must-Not-Be-Named".^[5] Algospeak is different from other types of netspeak in that its primary purpose is to avoid censorship, rather than to create a communal identity, though it may still be used for such end.^[1]

Causes and motivations

Many social media platforms rely on automated content moderation systems to enforce their guidelines, which are often not determined by users themselves.^[2] TikTok in particular uses artificial intelligence to proactively moderate content, in addition to responding to user reports and using human moderators. In colloquial usage, such AIs are called "algorithms" or "bots". TikTok has faced criticism for their unequal enforcement on topics such as LGBTQ people and obesity. This led to a perception that social media moderation is contradictory and inconsistent.^[1]

Between July and September 2024, TikTok reported removing 150 million videos, 120 million of which were flagged by automated systems.^[6] In addition, AI may miss important context; for example, communities who aid people who struggle with self-harm, suicidal thoughts, or past sexual violence may inadvertently get caught in automated moderation.^[7][8][1] TikTok users have used algospeak to discuss and provide support to those who self-harm.^[9] An interview with nineteen TikTok creators revealed that they felt TikTok's moderation lacked contextual understanding, appeared random, was often inaccurate, and exhibited bias against marginalized communities.^[1]

Algospeak is also used in communities promoting harmful behaviors. Anti-vaccination Facebook groups began renaming themselves to “dance party” or “dinner party” to avoid being flagged for misinformation. Likewise, communities that encourage the eating disorder anorexia nervosa have been employing algospeak.^[10] Euphemisms like "cheese pizza" and "touch the ceiling" are used to promote child sexual abuse material (CSAM).^[11]

On TikTok, moderation decisions can result in consequences such as account bans and deletion or delisting of videos from the main video discovery page, called the "For You" page. In response, a TikTok spokeswoman told The New York Times that the users' fears are misplaced, saying that many popular videos discuss sex-adjacent topics.^[12]

Methods

Algospeak uses techniques akin to those used in Aesopian language to conceal the intended meaning from automated content filters, while being understandable to human readers. One such method draws from leetspeak, where letters are replaced with lookalike characters (e.g. $3X for sex).^[3] Other similar adoption of obfuscated speech include Cockney rhyming slang and Polari, which were formerly used by London gangs and British gay men respectively.^[7] However, unlike other forms of obfuscated speech, the global reach of social media has allowed the language to spread beyond local settings.^[3]

Another method is where certain words may be censored, or in the case of auditory media, cut off or bleeped,^[a] e.g., s*icide instead of suicide. A third method involves "pseudo-substitution", where an item is censored in one form, while it is present in another form at the same time, as used in videos.^[13] Some methods involve intersemiotic translation, where non-linguistic signs are interpreted linguistically, in addition to further obfuscation. For example, the emoji "🌽" signifies pornography by means of 🌽 → corn → porn.^[3]

In an interview study, most creators that were interviewed suspected TikTok's automated moderation was scanning the audio as well, leading them to also use algospeak terms in speech. Some also label sensitive images with innocuous captions using algospeak, such as captioning a scantily-dressed body as "fake body".^[1] The use of gestures and emojis are common in algospeak, showing that it is not limited to written communication.^[14]

A 2022 poll showed that nearly a third of American social media users reported using "emojis or alternative phrases" to subvert content moderation.^[11]

Impact and detection

Algospeak can lead to misunderstandings. A high-profile incident occurred when Italian actress Julia Fox made a seemingly unsympathetic comment on a TikTok post mentioning "mascara", not knowing its obfuscated meaning of sexual assault. Fox later apologized for her comment.^[7][15] In an interview study, creators shared that the evolving nature of content moderation pressures them to constantly innovate their use of algospeak, which makes them feel less authentic.^[14]

A 2024 study showed that GPT-4, a large language model, can often identify and decipher algospeak, especially with example sentences.^[16] Another study shows that sentiment analysis models often rate negative comments incorporating simple letter–number substitution and extraneous hyphenation more positively.^[17]

Examples

According to New York Times:^[12]

• le dollar bean – lesbian, as derived from the written form Le$bian
• panini, panoramic – a pandemic, especially the COVID-19 pandemic^[18]
• accountant – sex worker
• (to) unalive, unalived – to kill; killed, dead.^[19]
• cornucopia – homophobia
• leg booty – the LGBTQ+ community
• nip nops – nipples
• seggs – sex

Other examples:^[20][1][21]

• tism – autism spectrum conditions
• fork – fuck
• grape – rape
• Panda Express – pandemic
• yt – White/Caucasian people
• blink in lio – link in bio
• shmex – sex
• opposite of love – hatred
• camping – abortion
• sewer slide – suicide
• pew pew – firearm
• cheese pizza – child pornography
• ouid – weed
• acoustic – autistic
• restarted – retarded

See also

• Bronyspeak – Specialized vocabulary and slang of My Little Pony fansPages displaying short descriptions of redirect targets
• Chilling effect – Discouragement of exercising rights by threats of legal sanctions
• Cant (language) – Linguistic term for jargon of a group
• Enshittification – Systematic decline in online platform quality
• Koalang – Fictional language
• Newspeak – Fictional language in the novel "Nineteen Eighty-Four"
• Un-word of the year – German ironic award
• Internet censorship

Notes

1. Bleep censor is commonly implemented in the editing process of TV production, mostly for censoring out obscene language, rather than by YouTube and TikTok video authors themselves.