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

Indeed. The site's privacy policy doesn't appear to cover sending unpublished draft posts to OpenAI.

There have absolutely been times over the years when I have written posts that I have subsequently decided to delete before posting, often because I have concluded that they might contain too much information that might make me personally identifiable. It had not occurred to me that unpublished drafts might be shared with OpenAI or indeed with any other external company.

I think that we pretty much agree to our posts being sent anywhere for any purpose.ETA Sorry I see what you mean. I thought it only applied to published posts.

Yes after we post them not when we’re thinking about posting them.

As far as I can tell, the AI tool is activated when you click in the "enter your thread title" box. In other words, before you publish the thread.

I don't think either the terms of use or the privacy policy explain this clearly.

The terms of use do give Mumsnet broad rights to use content "submitted" through the website. I am not convinced that many site users would consider a draft post to have been "submitted" prior to actually posting though. This needs to be made much clearer.

If I paste the OP into a new thread and click suggested title it comes up with:

"Proposed structured response format for sensitive AIBU family support posts"

From what ChatGPT tells me, it is not impossible that openai could produce a completely irrelevant response to a prompt requesting a suggested title. So it is plausible that Justine's explanation is true, regardless of how unlikely it might seem. So I'm willing to accept that explanation at face value.

What troubles me is that MNHQ made up two completely spurious explanations before the third one was posted, and Justine appears to think this was justified in order to try to stop the inevitable speculation.

And of course the revelation that our posts are shared with openai before we have decided to publish them.

Blimey, this thread has been an interesting read. Shocking initial responses from HQ, clearly demonstrates that no attention was paid to the opening post. I have no idea re coding and I’m definitely not tech savvy, but the idea that AI developed that of its own accord seems very … odd.

It’s not impossible but surely there would be an outcry from MNHQ and a big ole investigation if that’s what happened given the potential reputational impact on MN if users start thinking posts are written by AI?

The very fact that Justine responded herself speaks to the seriousness of the issue but strangely the content of her response doesn’t suggest that MNHQ considers it to be all that serious. I mean, isn’t this potentially existential-level stuff for MN?

I agree that the initial responses were shocking.

First the glib and frankly patronising dismissal of any concerns - "the prompt may look complicated" but all it's asking for is for AI to suggest a title. Well, no, we can all read...

Then the confident claim that it was nothing to do with MN and must have come from @AmIAComplete's own device.

Not having a go at the individual moderator as Justine made it clear that the moderator in question was acting on advice. But the poor quality response to the very reasonable concerns expressed by site users is surely unacceptable?

And I do acknowledge that Justine has apologised for the lack of thoroughness in the initial response, but at the same time, she appears to justify the fabricated explanations by implying that it is a "balancing act" between responding quickly and responding truthfully because MN is worried about encouraging speculation.

I think this tells us quite a lot about the culture at MNHQ. Instead of taking site user concerns seriously and engaging with them openly and honestly, they are more concerned about quickly shutting down any conversations that might have the potential to damage the site's reputation.

What they don't seem to have under is that the failed attempts to dismiss concerns do far more damage to users' trust in the site than anything.

https://www.mumsnet.com/talk/site_stuff/5122770-why-were-taking-legal-action-against-open-ai-and-other-scrapers

Lol. A few years ago you posted this and now you've admitted that you're actually sending peoples posts to OpenAi willingly

I agree. Justine was wheeled out to respond to the issue but then seemed to downplay it as if it was no big deal.

But it is a big deal.

MN's potential use of AI to generate fake posts is a massive deal - whether it happened or not, because there is enough evidence to at least make people reasonably suspicious that this may be what's happening.

MNHQ's casual admission that they share posts with openAI prior to publication is also quite a big deal imo.

And so is the casual glossing over of the two spurious "explanations" offered earlier in the thread before we got Justine's version, as if it is quite normal to just post any old crap to keep people quiet and prevent them from speculating.

I really hope that the media picks this thread up as MNHQ still has some questions to answer.

Our tech team have checked the tool and its logs, and they're confident it didn't come from us. Our best guess is that something else was running on the device while the post was being written, such as a browser extension, keyboard app or writing assistant.

I have nothing like this. No extension, no keyboard app or writing assistant.

I use Chrome on Android.

I am a prolific poster (name changed as I didn't want this true event linked to my other posts )

I have not made this up- what I posted in the first post is exactly what was suggested by the MN title option.

To be fair, they offered to licence the content to OpenAI, so I don't think they were ever bothered about sharing our data with large language modules, they just wanted to be able to profit from doing so.

By posting on the site, we consent to letting MN pretty do what they want with the data, but I had not been aware that I was consenting to my data being shared before I had even posted it.

I wonder if it was a rather dubious way of settling with Open AI.
Ie "you agree to stop scraping everything and in return we agree to push loads of our users to send you their posts under the guise of "helping" them formulate a thread title "

Justine has already backtracked on this, OP. They know it's nothing to do with you or your browser, though they haven't really explained why they ever tried to suggest that this was the case.

I've read further now and note Justine's response - I am certainly glad they found the error was there end and not mine.

As always I'm impressed by the wealth of knowledge of MN posters.

Thanks, I'm not normally such a rookie, but I was cross they thought it was me!.

Don't blame you!

I'm not sure whether I'm more shocked by OpenAI coming up, seemingly unprompted, with a schema for writing posts, or by MN passing unpublished posts to a third party, or by the way MNHQ have responded. I'm tempted to start an AIBU but that would be a TAAT!

What we can be sure of is that OpenAI is taking in every single draft post on MN and using it to train on.

Your unique writing style, the details of your personal life, your worries and hopes and passions - all go to train OpenAI in its quest to take over all of our jobs and plagiarise humanity’s creative endeavours.

Obviously this is all online and so it happens anyway with LLMs scraping the internet, but the AI title suggestion tool makes it extremely direct.

@JustineMumsnet how do I opt out of this processing?

I agree with PP @JustineMumsnet - I really don't think we need AI to create thread titles for us. I've never understood why we needed it, given that we're a discussion forum and not just a like/dislike sort-of platform.

It seems like using a massive, over-sized sledgehammer to crack a very small nut.

I don't understand why we need AI generated thread titles either.

If someone makes a decision to start a thread, the least I would expect is that they come up with their own thread title.

@biwi I don’t understand how we opt out of it, given that it’s automatic? That means our personal private data is being sent to OpenAI when we haven’t actually uploaded anything to Mumsnet.. we haven’t actually created a thread.

I’ve posted some really personal stuff here and I wouldn’t have done that if I’d know Mumsnet were directly sharing it with open ai.

so I want to know how do we opt out and how can I remove the information that has already been shared with OpenAI without my consent?