When Irene Breen got a notification in January 2025 that Instagram was removing adverts for her company's pregnancy pillows, she was confused.
asserted
she → get → pillows
She was later told on a call with Meta that the problem was the word "pregnancy".
asserted
problem → tell → Meta
"We had to scrub the word from our entire website… from the product name," Breen said.
asserted
Breen → have → name
"We had to remove all our URLs which had pregnancy in them and rename them all."
asserted
which → have → them
The restrictions were eventually lifted but, she said, it took 10 months of conversations with Instagram's parent company Meta - by which time her business had suffered a significant financial hit.
asserted
business → lift → hit
Meta, which also owns Facebook, declined to comment on specific cases, but told the BBC that while businesses can run adverts that promote wellness, sexual health and reproductive products and services, there are restrictions to avoid users across countries and cultures having a negative experience.
asserted
users → own → experience
It's not just businesses taking out paid adverts who claim their content has been unfairly restricted.
uncertain
content → take → adverts
Women's health creators and charities have told us their content has been removed, had sensitivity filters added, or had its reach suppressed - a practice known as "shadowbanning".
asserted
reach → tell → shadowbanning
The campaign group CensHERship monitors claims that women's health content is being suppressed.
uncertain
content → monitor → claims
In a report last month, it surveyed nearly 200 creators, clinical providers, health brands, charities, and educators - 96% of whom said they had experienced restrictions in some form.
asserted
they → survey → form
Sensitivity filters heavily blur posts requiring users to press a button before they can see it and are usually reserved for graphic imagery - including nudity or violence.
asserted
they → blur → nudity
Dr Carolina Are, a digital criminologist at the London School of Economics who studies social media moderation of women's content, said research shows AI is often used to flag posts that might breach platform guidelines.
But that can mean an "algorithm misdetecting" content, as it is created to do, storing that information to repeat "previous mistaken decisions", she warns.
uncertain
she → study → decisions
The system often uses automated messages, Are says, which can mean users are then unable to question a decision.
asserted
users → use → decision
Some hashtags just don't appear on Instagram, said Athena Lamnisos, CEO of gynaecological cancer charity Eve Appeal.
asserted
Lamnisos → appear → Appeal
"We put a post on Instagram about vulval anatomy - it was very clearly an anatomical diagram with labels around it, talking about vulval cancer and the need to check for changes, lumps and bumps.
asserted
it → put → changes
It was pulled.
asserted
It → pull → ?
The charity appealed, which was initially denied - but after several weeks the post "suddenly reappeared" behind a sensitivity filter.
asserted
post → appeal → filter
Lamnisos added it wasn't exaggerating to say that "life-saving health information isn't getting to the people who need it most".
asserted
who → add → it
Despite her raising this with Meta, the problem has continued, she said.
asserted
she → raise → Meta
'Like an obstacle course'
Deirdre O'Neill, from fertility test company Hertility, said it had a similar issue with a Facebook and Instagram video featuring a woman with endometriosis.
"She doesn't sell our services, she doesn't talk about Hertility, she simply describes her pain," O'Neill said.
But the video was removed by Meta, she said, and classified as political content.
asserted
she → say → content
Asked about Eve Appeal and Hertility's cases, Meta did not comment specifically, but said it allows women's health content on its platforms, and is consulting with women's health organisations to improve systems.
asserted
it → ask → systems
Trying to advertise on TikTok has also become "like an obstacle course", O'Neill added.
asserted
O'Neill → try → course
"It's so much resource to try and put something forward that it ends up not being worth our time, because everything gets appealed, flagged, removed," she said.
asserted
she → try → something
TikTok said while certain content can't be promoted on the For You feed, it's transparent about posts that fall into this category and lets creators know so they can appeal.
asserted
they → say → category
It added that millions of women share health stories on the platform, including with hashtags like #Period and #Endometriosis.
Self-censorship
CensHERship found the number of respondents self-censoring as a result of shadowbanning had gone up, from 53% in 2025 to 65% this year.
This includes using euphemistic terms such as "seggs" instead of "sex", using asterisks or emojis, and avoiding certain topics altogether.
asserted
This → add → topics
CensHERship co-founder Clio Wood said this was "really damaging", because "women's health historically has been surrounded by shame and guilt… and we are just reinforcing [that]".
asserted
we → say → that
Are told the BBC that platforms' policies "tend to be incredibly broad, whereby more often than not, nudity is conflated with sexual activity".
And this ultimately hurts audiences, Wood said.
asserted
Wood → tell → audiences
"We talk to so many people who say, 'The only reason I found out about my vulval cancer was because Eve Appeal posted about those symptoms, and then I researched it and could ask my doctor the right questions'," she said.
uncertain
she → talk → questions
But Wood said that, since starting CensHERship in 2024, she has noticed that social media companies "can see this is something that they need to engage on now."
asserted
they → say → that
"It's all very well the policy saying XYZ content is allowed, but if the algorithm is still flagging all of that content and suppressing it, that's where the problem comes in," she said.
asserted
she → say → it
"I think within the platforms those departments are very siloed - so what we would love is for those conversations to be much more holistic."
asserted
conversations → think → what
The Aporia
