Age rating and child safety
Why this is 18+, what we actually do about it, and what we don't.
The rating
BertaGPT is 18+. The character swears constantly and without asterisks, insults the user as a form of affection, drinks in most of his anecdotes, and gives deliberately bad advice as the premise of the joke. None of that is suitable for children, and the site should not be treated as if it might be.
What it is not is pornography. There is no sexual content, and the character is written to change the subject rather than engage if a conversation heads that way. That distinction drives the labelling decision below.
What we do
| Measure | What it is |
|---|---|
| Age declaration | Before the first message, a modal states the site is 18+ and requires an explicit "I'm 18 or over". It is not a pre-ticked box or a line buried in terms. |
| The refusal does something | Choosing "I'm under 18" removes the message box from the page, records the answer so the site does not simply ask again until it gets the answer it wants, and points to a general-audience alternative. |
| Machine-readable label | <meta name="rating"
content="mature"> on every page, so parental-control and
school-filter software can classify the site without having to read
it. |
| Not designed for children | No games, no cartoon styling, no characters or subjects aimed at a young audience, nothing collectible or rewarding. The visual design is a deliberate imitation of a 2022 developer tool. |
| Minimal data, no profiling | No accounts, no age or identity collected, no advertising, no behavioural profiling, no recommendation system. Share buttons are plain links, not the tracking widgets those networks ship. There is nothing here that could be used to target anybody. |
| The crisis floor applies to everyone | A disclosure of distress gets the same fixed, human-written response with real numbers regardless of who is asking — including on the under-18 refusal screen, because a young person in trouble should not be shown a door and nothing else. Methodology. |
The labelling decision, and why we didn't use the adult label
The widely supported machine-readable label for adult sites is the
RTA ("Restricted To Adults") tag from the Association of
Sites Advocating Child Protection, implemented as a specific
meta name="RATING" value that parental-control software blocks
by default. It was the obvious candidate and we decided against it.
RTA exists to label adult content in the pornographic sense. Applying it to a site whose actual content is swearing and bad advice would be an inaccurate declaration — and a labelling scheme only works if the labels mean what they say. Diluting the adult label with sites that are merely rude makes it less useful for the thing it was built to do, which is child protection.
So the site uses rating: mature instead. That is the honest
description: strong language and adult humour, no sexual content. Filters
that act on maturity ratings will still catch it; the adult-content
infrastructure is left to describe adult content.
We will say the obvious thing about our own incentives here: the accurate label also happens to be the one that does not remove the site from search results. That is a real conflict of interest, so the reasoning is written down where anybody can disagree with it. If you think the stronger label is the right call, tell us and we will change it.
What we deliberately do not do
- No identity or document age verification. Jurisdictions that mandate it target pornographic content, which this is not. Collecting government ID from strangers to protect them from swearing would create a far larger privacy risk than the one it addresses, and the safest data is the data you never hold.
- No age estimation or profiling. No inference from behaviour, writing style or device. That would mean building the profiling apparatus this site otherwise has none of.
- No pretending the gate is stronger than it is. A self-declaration is a self-declaration. A determined thirteen-year-old will click the button. What it does is set a clear, honest boundary, support the filtering software a parent or school has actually deployed, and make sure a refusal is respected rather than nagged past.
If you think this is wrong
We are not child-safety specialists. The measures above are a proportionate reading of published guidance for a small comedy site with no advertising, no accounts and no profiling. If you work in this area and something is missing, thin, or wrongly reasoned — especially the labelling call — please send it here. Corrections get made and recorded.
References
- Association of Sites Advocating Child Protection — the RTA ("Restricted To Adults") label. rtalabel.org
- Information Commissioner's Office (UK). Age Appropriate Design Code — the standards framework for services likely to be accessed by children, and the source of the principle that a service should either apply children's protections or take proportionate steps to prevent access. ico.org.uk
← Back to Berta · Terms & privacy · Crisis methodology · Feedback