Heroines, not heroin: charity’s Facebook web page returns after AI flagged it for medicine | Photography

When the UK charity Hundred Heroines had its Facebook group taken down it was accompanied by a message from the social media firm that merely mentioned the web page “goes against our community standards on drugs”.

Now, after greater than a month of interesting, the pictures charity is celebrating the reinstatement of its group after the tech firm’s AI instruments mistook it for an organisation selling the class-A opioid heroin.

The Gloucestershire-based organisation, which celebrates feminine photographers, has had its Facebook group taken down twice in 2025 for obvious breaches of neighborhood tips associated to the promotion of medication.

The newest takedown got here in September. After a second enchantment in 12 months, the Hundred Heroines: Women in Photography Today web page was restored with no rationalization or apology final week.

The charity’s founder and former president of the Royal Photographic Society, Dr Del Barrett, mentioned the choice had had a “devastating” impact on an organisation that depends on Facebook for attracting guests.

“AI technology picks up the word heroin without an ‘e’, so we get banned for breaching community guidelines,” she mentioned. “Then no matter what you do, you can’t get hold of anyone and it really affects us because we rely on Facebook to get our local audience.”

Founded in 2020, the charity has a bodily area in Nailsworth, close to Stroud, with about 8,000 items in its collection that focuses on the work of feminine photographers and spans the historical past of the artwork type.

In 2024, Meta elevated its vigilance of teams associated to medicine in gentle of the opioid disaster within the US, the place 80,000 individuals died after overdoses final 12 months.

Meta says shopping for and promoting medicine is strictly prohibited on its platforms, and claims to have “robust measures” in place to detect and take away such content material.

In a statement on its website, Meta says: “We recognise the significance of the drug crisis and are committed to using our platforms to keep people safe … and strict enforcement of our community standards.”

But when its software program incorrectly identifies teams that breach its requirements, the result could be Kafkaesque, in keeping with customers, with suggestions types typically the one approach to flag errors.

Meta says AI instruments are “central to [its] content review process”, including that “AI can detect and remove content that goes against our community standards before anyone reports it”.

Sometimes the know-how flags content material for its “human review teams”, though Barrett mentioned when Hundred Heroines complained they’d no human interplay.

“We thought, ‘should we change our name?’ But why should we? Why have we got to mess with our brand just because of Facebook?”, mentioned Barrett, who estimates that about 75% of Hundred Heroines’ guests come by way of Facebook.

She added: “It sort of verges on scary and laughable. You think these bots are running the world and they can’t tell the difference between a woman and an opioid. Heaven help us.”

Earlier this 12 months Meta was closely criticised over the mass banning or suspension of accounts on Facebook and Instagram.

Users blamed its AI moderation instruments for the inaccurate bans, however whereas the tech firm acknowledged a “technical error” affecting Facebook Groups, it denied a rise in incorrect enforcement of its guidelines throughout its platforms.

Meta mentioned it was fixing the problem that emerged in the summertime after teams, together with one which shared memes about bugs, have been allegedly informed they didn’t observe requirements on “dangerous organisations or individuals”.

Meta has been approached for remark.