The other side of content moderation: journalists must decode ‘algospeak’ for accuracy

PICTURE: RDNE Stock project/Pexels

Iris Pase

International Journalists’ Network

Skibidi toilet. Fruity. Sneaky link. If you’re older than 25, TikTok might feel like a jarring plunge into a world where you’re no longer ‘cool’ and the kids are speaking a language you don’t understand. While it’s tempting to dismiss these new linguistic quirks as just another fleeting trend, there’s more at play beneath the surface.

Beyond a platform for viral dances and teenage slang, TikTok is a testing ground for a novel form of communication known as ‘algospeak’. This coded language, born from the need to circumvent the platform’s unpredictable moderation, is reshaping how information is exchanged and understood online.

For journalists, staying abreast of this evolving language isn’t merely a matter of keeping up with the times – it’s essential for accurate reporting in the digital age.

What is algospeak?

Algospeak refers to the growing lexicon of alternative spellings, symbols and euphemisms used by creators to evade algorithmic censors on platforms like TikTok and Instagram. The term itself is a mash-up of ‘algorithm’ and ‘speak’, highlighting its origin as a form of coded language designed to outwit automated moderation systems.

As social media companies deploy increasingly sophisticated moderation tools, users – especially those discussing controversial or sensitive topics – are forced to develop creative workarounds to avoid having their content flagged, restricted or removed

Sex educator Evie Plumb knows these tools all too well. Last July, she lost her TikTok account after being repeatedly flagged by the platform’s content moderation system. Her work focuses on sex and sexual health education, which, while not explicitly banned, often triggers TikTok’s moderation processes.

‘I’ve had countless videos flagged or taken down,’ Plumb said. ‘I’d speak about completely innocent topics – things that are essential for people’s health and wellbeing – and they’d get flagged.’ In response, she adapted by using algospeak: ‘I started saying things like ‘seggs’ instead of ‘sex’ or censoring body parts.’

Plumb’s experience is far from unique. The rise of algospeak has turned platforms like TikTok and Instagram into arenas where creators strive to outsmart algorithms that don’t understand context and lack the nuance to distinguish between harmful and educational or journalistic content.

The evolution of algospeak

The emergence of algospeak reflects a growing tension between content creators and platform algorithms. Dr. Daniel Klug, co-author of one of the first academic studies on the phenomenon, explained that this coded language is constantly evolving.

‘The platform learns,’ said Klug. ‘If you spell “boobs” as “B00BS”, it won’t take long for TikTok to catch on, so people move to new codes, images or even gestures to communicate the same thing.’

For many communities, particularly marginalised ones, algospeak has become a vital tool for survival in a digital space where their voices are often suppressed. ‘Algorithms are not much different from society in general. Marginalised communities often get suppressed, pushed aside or deemed less important,’ Klug said.

Just as creators find ways around the algorithm, the platform adapts. As Plumb pointed out, ‘I’m starting to wonder, at what point will I not be able to use even censored words? It’s getting harder and harder – soon, I’ll have to communicate in ways that will feel like a riddle to anyone reading’.

Censoring language can make content less accessible, Plumb shared, particularly for people with learning disabilities or neurodivergent individuals, creating barriers to understanding and engagement.

Why journalists should pay attention

Algospeak is not just a quirky internet phenomenon – it’s a vital clue to understanding how sensitive information is communicated online. As moderation policies tighten, journalists must become adept at decoding the shifting lexicon or risk overlooking stories that matter

Erika Marzano, audience development manager at Deutsche Welle (DW), underscored this point. “We’ve learned that if we want our content to survive on TikTok, we have to adapt,” she said.

As a news organisation operating in 32 languages, DW frequently tackles sensitive topics – terrorism, drugs, LGBTQ+ issues – that don’t always fare well under TikTok’s moderation filters. ‘Sometimes we’ll publish a video about drug policy or extremism, and it gets flagged almost immediately,’ Marzano said.

To circumvent algorithmic censorship while maintaining clarity, and to keep stories accessible, her team has adopted algospeak, primarily using symbols, numbers and asterisks. Marzano explained: ‘We avoid euphemisms or alternative words because they can be very specific to a language, culture or group.’

‘For example, TikTok groups might say “corn” instead of “porn” or “accountant” instead of “sex worker”.’

This approach helps prevent misunderstandings that might arise from ambiguous euphemisms.

The challenge for newsrooms isn’t just avoiding takedowns; it’s ensuring the audience can still access important stories. ‘We’re not self-censoring,’ Marzano clarified, ‘but we are finding ways to communicate within the constraints of the platform’.

For journalists covering digital culture, this is where the stakes rise. Reporting on issues like sex education, drug policy or activism without understanding how these conversations are being shaped by platform rules risks not only missing the story but also misrepresenting it. If a community has shifted its language to evade moderation, reporters need to follow that shift to accurately capture the narrative.

The limits of algorithmic moderation

The rapid evolution of algospeak exposes the core weakness of algorithmic moderation: its inability to understand context. ‘The problem is that algorithms don’t understand context. They can’t tell if I’m being sarcastic or not,’ Klug said. ‘They see a word, label it as ‘bad’ and suppress the content, even if it’s being used in an educational or non-threatening way.’

This limitation becomes especially problematic for content that touches on sensitive or controversial topics. DW, for instance, has found that its coverage of historical events like Nazi-era Germany or terrorism often triggers automated flags. ‘We’re not promoting these ideologies, but the algorithm doesn’t differentiate,’ Marzano said. ‘It sees a particular picture or word and takes it down.’

Algospeak and the future of journalism

As social media platforms evolve, so does the landscape for journalism. Algospeak, with its shifting codes and euphemisms, highlights a pressing reality: navigating digital communication effectively requires more than just understanding the current language. It necessitates adaptability in content dissemination and business models

Marzano stressed the importance of not relying solely on one platform. ‘The reason we say we are not self-censoring is because the topics we don’t cover on TikTok are covered elsewhere. We have our website, our streaming platform, our podcast and our newsletter.’

The need for adaptability can also impact monetization strategies. As platforms implement new algorithms and policies, revenue streams linked to these platforms can be unpredictable. By embracing a multi-channel strategy including podcasts, newsletters, and video streaming, news organisations can better manage financial risks and capitalize on multiple revenue opportunities.

As algorithms and platform policies continue to evolve, journalists must remain agile, leveraging a diversity of platforms and adapting their strategies to ensure their stories reach the intended audiences. Understanding and navigating algospeak is not just a matter of keeping up with digital trends; it’s essential for maintaining journalistic integrity and ensuring that important stories are accurately reported and widely accessible.

By embracing a diverse approach and staying attuned to the nuances of digital communication, news organisations can enhance their resilience and financial stability in an increasingly complex digital landscape.

• Read the original story here
• Iris Pase is a freelance journalist based in Glasgow