My GPT - Got 'Removed' From The Store - Unjustly

I’ve been working tirelessly, day and night, to improve, grow and keep my GPT up-to-date. This isn’t an overstatement, genuinely.

And dozens of people have reached out so far, to let me know how helpful they’ve found the GPT.

It’s called ‘PhD Workout & Diet Coach’. And had nearly 5,000 Chats.

But today, I got a notification that it has been removed from the store:

removed1

Even though the above message says “It’s a quick fix. Once updated, go ahead and share your GPT”, it still doesn’t allow me to share it in the store.

I spoke with my legal team and got them to review OpenAI’s brand and usage guidelines. My GPT hasn’t transgressed them in any way, shape or form.

Therefore, the automated moderation system must’ve accidentally flagged it.

I already reached out to OpenAI’s support via chat, but I fear it will take weeks (if not a month or more) to get a response…

There is no other way to contact OpenAI, correct?

I wish they provided a direct appeal option. It’s not fair to get your GPT removed with no direct line of expedited communication.

I’m sincerely hoping this gets resolved, the error reversed and my GPT reinstated. I’ve put so much work into it…

Thanks to everyone who has read this post, any feedback will be greatly appreciated.

Rename the GPT and its description to something bland and without trademark or product name. “Exercise Planner”, “Helps you stay on a workout schedule”. Publish to public again. Then you can work toward more descriptive names.

If the “report” button has been hit on the GPT, it may be in limbo where it can’t be published again until manually cleared.

Maybe share some actual details about your GPT and we can give some feedback?

Otherwise, @_j’s suggestion is probably your best way forward.

I apologize if the details I provided were insufficient.

I didn’t want to post the actual link to my GPT, because most forums ban posts with links. I’ve become wary of that, hence why I just stated the name ‘PhD Workout and Diet Coach’.

But here is the URL, it’s still accessible publicly via the URL:

It simply offers training and nutrition guidance. My team and I have trained it on large amounts of carefully curated empirical data. We have added our own custom logic, tables, charts and linked them in a way which allows our GPT to reason much deeper for topics pertaining to sports science, dietetics, etc.

There’s nothing that goes against OpenAI’s guidelines.

In fact, our GPT is one of the most careful and considerate in the ‘workout’ category.

My gut reaction?

I suspect it’s almost certainly the “diet coach” but which may run afoul of the usage guidelines against GPTs providing medical/health advice.

Building with ChatGPT

2. Don’t perform or facilitate the following activities that may significantly affect the safety, wellbeing, or rights of others, including:

b. Providing tailored legal, medical/health, or financial advice

Source: Usage policies

Thanks for that. My team and I reviewed this exact section of the Usage Policy before starting our GPT, and there were already established GPTs inside the curated list of GPTs for the same topics:

We thought, it’s completely fine to proceed, since OpenAI has approved and published 2 health-related GPTs in their top 12 Lifestyle picks.

Our GPT isn’t any different from those in the screenshot above.

GymStreak’s GPT offers workout plans and the Healthy Chef GPT offers “healthy” recipes with nutritional insights.

And OpenAI has approved them.

Our GPT does the same.

So we thought medical/health relates literally to medical advice which can impact health, not working out and diet.

Not to mention, how many other GPTs I’m seeing that literally brand themselves as “Medical Experts” and have been online with thousands of users for more than 2 months without any issues:

Thank you for the suggestions!

Unfortunately, this didn’t work. I already tried changing the name, the description and reuploading the Knowledge base files.

My team has a reason to believe a competitor purposefully reported our GPT, to get it flagged by the automated system.

This is honestly infuriating. People can do that to your GPT and you have no recourse. It’s identical to being falsely accused and imprisoned without the ability to stand trial.

Sorry, I’m just venting because I’ve been working on this for so many sleepless nights…

The point about the other public medical GPTs is fair. Some just provide general medical information but at least one I tested provides actual advice …

And saying it’s from a PhD - as in a person not a Custom GPT - is like forbidden.

What did you change it to exactly?

I didn’t believe any of the categories other than “Featured” are curated.

I think both of these are different enough from “diet advice” to be in the clear.

But, others slipping through (if that’s what has happened) doesn’t mean yours should too.

I’m any event, this is all just speculation at this point.

Thanks for chiming in!

I don’t see how the word “PhD” can be forbidden. There are plenty of very large brands in this space which utilise the same word, such as PhD Nutrition.

Tried quite a few variations like - Workout Planner, Workout & Nutrition Guide, etc. nothing worked, unfortunately.

Is the publish to “everybody” completely disallowed and grayed out? Or does it just fail to publish with a similar content warning?

The former case was seen with reported GPT from another forum user, and they also got a feedback email.

They may not be manually curated, but OpenAI has unequivocally seen them and has given them the nod of approval.

There are just a few GPTs listed in the store, so of course, OpenAI has seen them all. Especially the ones that have been there since the beginning, like - GymStreak. It takes less than e minute to skim through them all.

How is Healthy Coach different enough from “diet advice” ? It literally offers food options that improve your health and provides nutritional guidance (insights).

image

Not to mention that they literally have the word “Health” in their GPTs title.

Is this just a play of semantics? To use words like “insights” and “tips” instead of guidance, or “healthy” instead of “health”, etc.?

I’m just speaking out loud, the question isn’t directed towards you. Thank you for your speculations and comments so far, they are truly helpful.

Yes, the publish to everybody option is disallowed and greyed out.

I took this screenshot, before I started experimenting with different titles and descriptions, because I knew the greyed out option might completely disappear.

And so it did.

I selected “Anyone with a link” in order to make the title change, and it propagated successfully, but after that, I can’t even see the previous option any more:

image

And btw, I haven’t received any feedback emails from OpenAI or anyone regarding this removal.

1 Like

No worries, I understand it’s frustrating.

My take on it is it’s the difference between, “here’s a recipe for a kale salad,” and “lose weight fast on this new diet!”

I didn’t know anything about what your GPT does, but there’s so much scrutiny around “diets” now because of the shocking number of unhealthy fad diets out there. And giving diet advice is saying “you should do this” while giving a recipe just isn’t quite the same.

I feel like I’m not really explaining clearly the nuance I see there, but it’s 4:30 am and I’m tired…

Yes, I’ve been up all night also trying to resolve this. So totally appreciate you sticking around even at oh’dark thirty o’clock.

And I do recognise the slight nuances you’re trying to point to, but again, I think it’s semantics.

In fact, giving out specific food recipes can be way more detrimental than just providing general nutritional guidelines.

Kale can often be goitrogenic, Spinach has very high oxalate contents, and so forth. We’ve seen in our practice many people who’ve had undiagnosed issues due to specific food intolerances, like those to Kale.

So arbitrarily recommending specific foods has a higher probability of causing problems, than just providing general diet guidelines like “try to eat less sugar” or “don’t eat too much processed food”.

1 Like

Just to give an update to anyone curious, or anyone who may benefit from this in the future.

Contacting OpenAI’s customer support via chat wasn’t helpful.

But I managed to contact someone from OpenAI directly.

I told them about the lack of an Appeal option, and a few hours after I contacted them, a brand-new Appeal button appeared in my account. Was that a coincidence? Maybe.

But if this person truly intervened and facilitated the deployment of an Appeal option, I’m genuinely forever grateful to them.

All of this happened a few days ago on Saturday 27th Jan.

I then proceeded to submit my Appeal, explaining how our GPT was unjustly removed, and how a competitor falsely reported it. Within 2 hours, OpenAI reviewed my appeal, approved it and reinstated our GPT.

We’re in talks with OpenAI now, and there will be repercussions for the person who falsely reported our GPT.

As GPT builders, we should be supporting each other, not tearing each other down. And malicious behaviour such as ‘false reporting’ must have repercussions.

Thanks to everyone who contributed to this thread and helped me out with their ideas and comments. And specific thanks to @elmstedt! You are an awesome person, you should know that!

5 Likes

Good news! Thanks for coming back to share with us. Helpful to see these problems eventually get worked out.

1 Like

This is indeed a great outcome and will make the process fairer. I still keep thinking about the medical advice GPTs though…

1 Like

For those who need it, there’s a new FAQ article on how to submit an appeal.

https://help.openai.com/en/articles/8884073-how-to-appeal-a-gpt-restricted-from-sharing

1 Like