Algospeak is the coded vocabulary TikTok creators use to say things the platform’s filters would otherwise catch. A creator writes “unalive” instead of “suicide,” or “penjamins” and “gardening” instead of “cannabis,” swapping the flagged word for one the system doesn’t recognize.
The workaround exists because TikTok’s moderation scans for exact words, not intent, and getting flagged carries real financial consequences for creators. Understanding how algospeak works, and why it keeps evolving, explains a lot about how language moves online today.
Why TikTok Censors So Aggressively, and Who Gets Hit Hardest
TikTok operates one of the most aggressive content moderation systems of any social media platform, and the financial stakes for creators make that pressure impossible to ignore. Creators in the top 7% of the platform earn around $200,000 annually, while the average sits closer to $20,000. When a single piece of flagged content can derail that income, self-censorship creators practice becomes less of a choice and more of a survival strategy.
The platform’s TikTok Community Guidelines prohibit an extensive list of topics, and the consequences for violations extend well beyond content removal. Shadow banning, where the For You Page algorithm quietly limits a video’s reach without notifying the creator, is the punishment most feared, because it is invisible. Your content stays up, but nobody sees it, and you may not realize why your numbers have collapsed.
How TikTok’s Algorithm Flags Content
Automated content filtering on TikTok works by scanning captions, on-screen text, and audio transcriptions for terms that appear on the platform’s banned or restricted lists. The system does not read context. It matches words against a database, which is why a video about suicide prevention can be flagged just as readily as one that glorifies it. This is the foundational logic that makes algospeak necessary: the only reliable way to avoid the filter is to use a word the filter does not recognize.
The detection net has grown wider over time. TikTok’s moderation systems have expanded beyond keyword matching to include visual recognition and hashtag analysis, meaning that internet language censorship workarounds need to be increasingly creative to function. Creators who relied on a single substitution word often found their reach collapsed once the platform updated its filter to include the new term.
Cannabis Content and Platform Moderation
Cannabis content has faced some of the steepest platform suppression of any legal consumer category on TikTok. Even in states where cannabis is fully legal, content discussing products, consumption, or medical use is routinely restricted or demonetized. Veriheal, as a licensed cannabis telehealth platform, has encountered this directly: content that meets every legal standard for the markets it serves can still be deprioritized by the For You Page algorithm because the underlying topic triggers moderation flags.
This creates an odd situation where a company operating entirely within state law is penalized by a platform’s federal-era content standards. Cannabis laws and platform policies vary by jurisdiction, and what is permissible in Colorado may still be restricted by TikTok’s global moderation framework. The result is that cannabis creators, educators, and healthcare providers have become some of the most practiced users of coded language online.
What Is Algospeak? The Linguistic Toolkit TikTok Creators Use
TikTok censorship euphemisms exist because creators need a way to discuss restricted topics without triggering automated filters. Algospeak is the name researchers and linguists have given to this system of coded substitutions, and it is more structured than it appears. When a creator says “unalive” instead of a clinical term for self-harm, or uses penjamin to mean cannabis, they are applying a consistent set of linguistic strategies, each targeting a different vulnerability in automated detection.
Understanding why TikTok users replace words with coded language like “unalive” or “penjamins” comes down to one simple fact: the platform’s filters match exact words, not intent. Change the word, and the filter misses it. The academic literature on this phenomenon, including a 2023 Indiana University study on algospeak strategies by Calhoun and Fawcett, identifies five documented techniques that creators draw on.
Phonetic Substitution
Phonetic substitution replaces a flagged word with one that sounds similar but is spelled differently. “Seggs” stands in for sex, and “le dollar bean” emerged as a spoken approximation of “lesbian” after straightforward spelling proved unreliable. The filter hears or reads a different word while the audience understands the intended meaning without any difficulty.
Morphological Manipulation
Morphological manipulation inserts symbols, numbers, or extra characters into the middle of a word to break the string a filter would match. Spellings like “le$bian” or “s3x” work by making the text unreadable to a keyword scanner while remaining entirely legible to a human reader. This technique became widespread early because it required no agreed vocabulary, any creator could apply it independently.
Euphemistic Replacement
Euphemistic replacement is the most creatively expressive of the five techniques, and it is where gardening as euphemism fits. Creators substitute a completely unrelated word or phrase that carries shared meaning within a community. “Gardening” refers to growing cannabis. “Corn” stands in for pornography. “Penjamins” describes cannabis in a way that required community familiarity to decode.
These substitutions function as a form of linguistic creativity online communities develop collectively, and they carry a layer of in-group signaling that the other techniques do not.
A Quick Reference, Common Algospeak Terms and What They Mean
These terms were coined to evade automated keyword detection, and the list evolves continuously as platforms update their moderation systems in response to each new wave of substitutions. Think of it as a living digital slang dictionary rather than a fixed vocabulary.
| Algospeak Term |
What It Replaces |
Why It Was Coined |
| Unalive |
Suicide / self-harm |
Flagged under mental health moderation policies |
| Seggs |
Sex |
Phonetic match avoids keyword filters |
| Penjamins |
Cannabis |
Euphemistic community term for plant and products |
| Gardening |
Growing cannabis |
Unrelated word with shared meaning in cannabis communities |
| Corn |
Pornography |
Sounds similar; visually unrelated |
| Le$bian / Le dollar bean |
Lesbian |
Morphological and phonetic workarounds for LGBTQ+ content suppression |
| Spicy eggplant |
Vibrator / sex toy |
Emoji-adjacent euphemism for adult product content |
When Algospeak Leaves the App: How Coded Language Enters Real Life
The most interesting dimension of algospeak’s spread beyond the platform is what happens once it escapes. Language evolution via social media has always moved terms from niche online contexts into general use, but algospeak follows an unusual path because it was never intended for general audiences in the first place. It was a workaround. Yet terms like “unalive” have migrated into mental health conversations far outside TikTok, used by people who have no idea the word was invented to dodge a filter.
Gen Z internet slang has always carried identity-signaling functions alongside its surface meanings, and algospeak adds an extra layer to that. Knowing what “penjamins” means, or using “corn” correctly in a sentence, marks you as someone who understands the coded landscape of the platform. The words become shibboleths: proof of fluency in a subculture. This is why algospeak terms spread even among people who have never experienced platform suppression themselves.
The question of whether this cannabis-based algospeak affects our speech in durable ways sits at the intersection of sociolinguistics and platform design. What the research on language shift suggests is that words introduced through high-frequency, emotionally resonant contexts tend to stick. TikTok’s scale, 170 million registered American accounts as of 2025, gives algospeak terms exactly that kind of exposure, repeated across billions of video impressions in contexts that carry real stakes for the people using them.
The offline spread also creates generational friction. Parents and educators encountering “unalive” for the first time sometimes interpret it as flippant, without recognizing that the word was coined precisely because creators wanted to discuss serious topics with care, not carelessly. The coded language online that began as a censorship workaround has become a genuine part of how a generation talks about difficult things.
Does Algospeak Actually Work?
The honest answer is: sometimes, and temporarily. Algospeak is effective at the moment of coinage, when the substitution is too new for the platform’s moderation systems to have catalogued it. But the effectiveness degrades quickly. TikTok publishes quarterly Transparency Reports documenting updates to its moderation systems, and those updates respond directly to the coded terms that have become widely used. “Unalive” was initially effective as a workaround. It has since been added to TikTok’s moderation lexicon, which means the evasion it offered is now gone.
This creates the arms-race dynamic that defines algospeak as a phenomenon. Creators coin new terms, platforms update filters to catch them, and creators adapt again. The cycle is structural, not accidental. Free speech on social media platforms is constrained not by a fixed set of rules but by a system that updates continuously, which means the vocabulary of evasion has to update continuously too.
For cannabis creators, the arms-race problem is compounded by the fact that content can be flagged on signals beyond text. Veriheal has observed that cannabis content using euphemistic language for products still attracted suppression when other signals, visual content showing recognizable products, hashtags with cannabis associations, or account history, triggered the filter independently. Algospeak addresses the text layer of moderation, but the moderation system has multiple layers, and a coded caption does not protect a video whose visuals make the subject obvious.
The result is that algospeak works best as one tool among several, not as a reliable solution on its own. Its real value may be less about reliably evading filters and more about the community-building and identity functions the language has taken on, functions that persist regardless of whether the words still fool a detection algorithm.
Final Thoughts
The coded vocabulary TikTok creators have built is not just a workaround. It is evidence of what communities do when institutional channels fail them: they invent new language, share it, and carry it beyond the original context until the words mean something larger than the problem they were designed to solve. Algospeak started as a patch for a broken system and became part of how a generation talks about sex, identity, cannabis, and mental health.
If you are a cannabis patient navigating questions about legal access and medical use, speak with a licensed Veriheal physician who can help you understand your options in your state.
Note: The content on this page is for informational purposes only and is not intended to be professional medical advice. Do not attempt to self-diagnose or prescribe treatment based on the information provided. Always consult a physician before making any decision on the treatment of a medical condition.
Note: Veriheal does not support illegally consuming therapeutic substances such as cannabis but acknowledges that it transpires because of the current illicit status, which we strive to change by advocating for research, legal access, and responsible consumption. Always consult a physician before attempting alternative therapies.
FAQ
What is algospeak?
Algospeak is a system of coded substitutions that social media creators use to avoid triggering automated content moderation filters. Instead of using flagged terms directly, creators replace them with phonetic approximations, symbol-altered spellings, or unrelated words that carry shared meaning within their community. The term covers any word or phrase coined specifically to evade platform detection systems like TikTok’s For You Page algorithm.
What does “gardening” mean as a TikTok euphemism?
On TikTok, “gardening” is used as a euphemism for growing cannabis. The term developed as a gardening-as-euphemism workaround because direct references to cannabis cultivation were subject to content moderation and platform suppression. Creators in cannabis communities adopted it as part of a broader coded vocabulary for discussing the plant and its products.
What does “penjamins” mean on TikTok?
“Penjamins” is an algospeak term used in cannabis communities on TikTok to refer to cannabis in general. It functions as a euphemistic replacement that allowed creators to discuss cannabis content without triggering TikTok’s automated keyword filters. Like most algospeak terms, its effectiveness as a workaround diminishes as the platform updates its moderation systems.
Why do TikTok users say “unalive” instead of other words?
“Unalive” is a TikTok censorship euphemism for suicide and self-harm, coined because direct language on those topics triggers automated content moderation even in educational or supportive contexts. Creators discussing mental health used the term to stay visible on the For You Page algorithm while still addressing serious subjects. It has since entered broader use outside TikTok, though the platform has now added it to its own moderation vocabulary.
Does algospeak actually prevent content from being suppressed?
Algospeak works inconsistently and temporarily. Substitution terms are effective when they are new, before the platform’s moderation systems have catalogued them, but TikTok updates its filters regularly and adds widely-used coded terms over time. For cannabis creators especially, text-based workarounds may not protect content when visual signals, hashtags, or account history independently trigger moderation.
Author, Share & Comments