Odd AI-generated thread title options in Site Stuff

[AmIAComplete]

AIBU Response Schema v1.0: Structured adjudication for interpersonal reasonableness posts about family support during a child’s serious illness, with special handling for ND context and empathy calibration. Use this schema to produce concise, practical verdicts and communication guidance without escalating stigma or family conflict unnecessarily. The assistant should assess whether the poster is being unreasonable, acknowledge the child’s medical context, weigh the brother’s expectations and possible ND-related rigidity, and give a clear, forum-style ruling plus brief actionable advice. Definitions: 'AIBU' = 'Am I Being Unreasonable?'; 'ND' = neurodivergent/neurodivergence; 'DD' = dear daughter. Safety/quality constraints: do not diagnose anyone; do not demean ND people; do not imply a sick child owes social performance; do not overstate certainty about motives; distinguish the child nephew’s needs from the adult brother’s message; prioritize compassion, boundaries, and de-escalation. Output should be parseable JSON matching this schema exactly, with no extra keys, comments, or prose outside JSON. In verdict, choose one of the enumerated values. In confidence, give 0..1. In core<span class="underline">judgment, give a short forum-style judgment. In reasoning, briefly explain balancing factors, including that serious illness and treatment commonly affect energy/mood; the brother may be stressed but his complaint is misplaced; ND can explain disappointment but not obligate the ill child to perform. In advice, include immediate emotional validation and one or two practical next steps (e.g., do not engage tonight, send a calm boundary later, loop in parents only if needed). In suggested</span>reply, provide a short message the poster could send to the brother, neutral but firm, avoiding argument. In tone<span class="underline">flags, indicate whether the response should emphasize empathy, boundaries, de-escalation, and whether to avoid legal/medical advice. In risk</span>notes, note any caution about the poster’s acute stress and potential family fallout. minor<span class="underline">child</span>considerations should explicitly state that children’s wellbeing takes priority over adult expectations in family gatherings when a child is newly seriously ill. nd_sensitivity should acknowledge ND respectfully while rejecting its use as a trump card over another child’s health needs. Keep outputs concise but humane, roughly the level of a supportive forum reply translated into JSON fields.

Weird shit on AI title options

But the AI worked on this post

Yep. Agree with all of the above.

Sorry, I didn't mean to suggest you were.

I asked ChatGPT if it thought MN had done anything illegal!!

It responded:

Has Mumsnet breached UK data protection law by sending users' unpublished drafts to OpenAI?

On the evidence currently available, I think Mumsnet has probably breached UK data protection law.

The strongest argument concerns Articles 5(1)(a) and 13 of UK GDPR, which require personal data to be processed fairly and transparently, with users properly informed about how their information is used.

There are several reasons for concern:

1. Unpublished drafts were sent to OpenAI. Mumsnet has acknowledged this, but its privacy policy describes OpenAI as helping it understand user discussions. That is not the same as informing users that their unpublished writing would be sent to an external AI system.

1. Legitimate interests may not provide sufficient justification. Mumsnet might argue that AI-generated titles improve its service. However, the ICO says organisations must consider whether people would reasonably expect their information to be used in this way. Sending unpublished drafts to OpenAI merely to suggest thread titles may be difficult to justify, particularly without adequate notification.

1. Sensitive personal information makes the issue more serious. Mumsnet users frequently discuss health, pregnancy, relationships and sexuality. Unpublished drafts could contain special-category personal data, which requires additional legal justification under Article 9 of UK GDPR. An unpublished draft cannot simply be treated as information someone has deliberately made public.

1. The feature operated automatically. Mumsnet's original announcement in March 2026 described AI-generated title suggestions appearing automatically, rather than being something users actively requested. This makes it harder to argue that users knowingly chose to send their drafts to OpenAI.

1. Transparency is a legal requirement, not an optional courtesy. The ICO requires organisations to explain how personal data will be used before processing begins. Simply including a general reference to OpenAI in a privacy policy may not satisfy this requirement.

Mumsnet's assurances that OpenAI was not permitted to use the drafts for model training do not resolve the central issue: whether Mumsnet was legally entitled to send the material to OpenAI in the first place.

A definitive conclusion would require examination of Mumsnet's data-processing agreements, its documented legal justification and exactly what information was transferred.

However, the available evidence raises substantial concerns about compliance with UK GDPR. An ICO investigation would be entirely warranted, and if the facts remain substantially as reported, I think a finding of unlawful processing is more likely than not.

The revelations only emerged in early October 2026. There has scarcely been time for a regulatory investigation, let alone a formal determination. The absence of an ICO finding at this stage should not be mistaken for evidence that Mumsnet acted lawfully.

No worries!

OpenAI policing itself! Is OpenAI maybe also on the hook for knowingly processing the data if it wasn’t GDPR compliant? It’s an interesting topic.

Another thing that is interesting about all this is that @JustineMumsnet does things like give evidence to parliamentary inquiries, based purportedly on what Mumsnet "users would be concerned about."

If this is based on analysis of AI-generated content, what on earth does this mean for authenticity and honesty and political processes? And whose / what's voices are heard and amplified?!

committees.parliament.uk/oralevidence/17712/html/

Does "the tool has been switched off" mean:

"We have disabled the channel that transmits data to OpenAI."

Or

"We've turned off the box that displays the OpenAI result."

?

Good question. I suspect option 2. Were still doing it but now you can't tell quite as obviously. I did laugh when it made the guardian Thursday quiz.

It would be GREAT if AI could be used to spot fake AI posts because they are super annoying.

I see from the Guardian article that MN users dropped 19% in 2025 - interesting correlation with my own experience of quality declining and the site being the victim of a lot of AI slop and fake rage bait stuff.

As a side note, devastated to not be quoted in the Guardian - exciting times @MrsBennetsPoorNervesAreBack 😆

Hmm. Very good question.

Still basking in the glory @Attenboroughsmistress!

Me too - going to add it to my CV 😂

AmIComplete whose commentary on Artificial Intelligence has been featured in The Guardian, driving national conversation on digital ethics...

Unfortunately, I suspect this will just become another thing in the rear-view mirror that most Mumsnetters won't care about for very long.

I feel like we used to have higher standards than this. People used to be held to account. Now any old explanation gets trotted out and, provided it's "truthy" enough, that's apparently the end of it.

Bah humbug.

I’m surprised no-one has taken Justine up on this offer to see the code.

Yes, I really hope not and that they are going to return and address this properly. Possibly things are going in in the background (inc with the Information Commissioner?)

I don't think it's okay that this is all being normalised.

a) People think they are talking about and genuinely helping and empathising with each other re super-sensitive subjects, and choosing what to publish. But their draft inner-workings are definitely being sent off without their knowledge. And, they are probably being responded to by AI, not the peers they think they are talking with.

b) Additionally, probably, AI generated content being presented in the political process, by Mumsnet, as representing real Mumsnet users

This is truly dystopian. It is a crunch point.

It's a good offer, and does need to be responded to by 'someone who knows about these things'.

You mean work for free, for a company that earns top dollar for the content already given them for free?

They should be able to hire people to sort this out. It isn't a dinner table operation any more.

External IT auditors should be brought in, with the results of the audit made public.

I read it as a general list of rules ( nice to know they have some I guess) plus a proforma for a responce to a particular AIBU question.

Not to sort it out for them, no. To check on them. For your own personal peace if mind (it would be a kindness to other users to share your opinion, but not an obligation).

@KatyMumsnet @JustineMumsnet Could I use this thread to ask you about some AI-related questions that are really unnerving me?

I have used multiple namechanges over the years decades to protect my privacy. I'm concerned that MN's own AI, and the functionality it may make available to its advertising clients, 'knows' that all of my usernames are me, on the basis of my single registration.

Clearly the tools that you use for site monitoring can see the identity of one person across multiple usernames, but will that functionality be expanded for other purposes, and does your data protection policy make that clear to users? What provisions do you have in place to protect the privacy generated by changing user-names, which has been a core convention of the site since its inception - the very basis on which we post so freely.

ETA: To clarify, what I'm worried about is an AI search function that can build a picture of us by aggregating all of our posts, under all names. Obviously that would be economically valuable for advertising purposes. It is already alarming how detailed and outing an AI search under just ONE username would be. If they blend all usernames they will have a picture that is both outing and revelatory of extremely private stuff.

I know you claim all data used for business purposes is anonymised, but what does 'anonymised' mean in very detailed terms? How vulnerable is it to mission creep and data protection incidents?

My most anxious questions are about the effects of deregistering. Obviously all my posts will remain, but will their connection to my email address and other registration data (and metadata?) be severed? I imagine that in order to prevent the return of banned posters, you do, legitimately, need to keep a record of their posts and registration. But does your data protection policy allow you to keep these details in perpetuity, and use them for business-related purposes?

GDPR requires you to ensure that you don't keep personal data longer than is necessary. I'd like to be reassured that ALL of the registration data you hold about me for site monitoring purposes would be deleted after, say, a year at most.

Absofucking lutely

I had considered asking our resident AI expert in my work if he'd like to take a look, he'd love a "mystery" like this.

But that would mean admitting I go on MN, and, to be frank, I'd rather slam my tit in a car door.

I’m deeply suspicious of the Penis Beaker story now.

Has this thread been hidden from ‘Active?”

Edited: No it hasn’t! Im getting paranoid now…