In advanced voice mode the system is constantly interrupting me saying “My rules does not allow me to speak about this” which renders the system unusable. Topic can be anything. Nothing controversial. This seems like some big bug.
Have you tried uninstalling the app? Maybe check the custom instructions, do you have anything there that might be triggering this?
Thanks. Not yet. Will try.
Same here!
The advanced audio mode constantly keeps flagging my input, with no reason. Even if I ask about very basic things, it keeps flagging my input, especially when the conversation becomes longer. At the very most basic stuff, with honestly nothing malicious whatsoever being talked about.
5 Likes
same here. constantly popup even talking daily life content. it happens more when we are using foreign languages.
i dont understand it is flagging my voice or the message it is about to speak.
I suspect it is my headphone mic issue, many environment noise that triggers the flagging.
2 Likes
I don’t really use advanced voice mode with ChatGPT but this isn’t completely surprising. I’ve seen similar aggressive flagging of o1s responses.
It’s not anything in your content it’s their safety model. They told me that for o1 it still needs a lot more tuning and I’m assuming same issue with Advanced Voice Mode. This is actually a really tricky problem
It’s tricky because you’re not going to notice if it lets through a response it shouldn’t have but you will immediately notice if it blocks a response it shouldn’t have. It’s a no-win scenario for the safety model as it can only ever be perceived of as having flaws.
1 Like
Well, you built the Karen AI! 
3 Likes
Same here. Prevents me talking about Bible.
This is definitely a bug that needs to be fixed.
1 Like
I’ve had the same exact problem ever since I got advanced voice mode. Nothing helps! ”The rules prohibit…” I discontinued my plus subscription, because I can’t bring myself to pay so much for a broken product. Please fix this! To date there’s sadly no competition for the advanced voice mode even though every big tech company is launching their own chatbots.
1 Like
Is AGI they said 
→ the model become sarcastic, he got feelings, you have to show some respect first 
Same problem here, talking either in Dutch or in English. Very frequent interruptions by another voice stating “My guidelines won’t let me talk about that” - really annoying! Is there any way to just disable this? Or just say nothing instead & just moving on to the next topic, without this annoying interruption? With this bug Advanced Voice mode now became practically unusable… Having the same chat via text presents absolutely no problems, so somehow the safety settings for Advanced Voice mode are put way too sensitive…
I do not get the “rules do not allow” message, but 90% of the conversations end up the AI interrupting me in the middle of a sentence giving me long winded answers about things I didn’t even ask yet. Then it loops to me trying to instruct the AI to listen to me until I have finished my question, to listen to me until it hears a key word for it to be OK to starts responding. Nothing helps.
Once I enter the interruption loop the AI will start and answering 2-3 seconds after hearing my voice, often answering to an imaginary question. Since the advanced voice rolled in, I have been exclusively using the “start recording, stop recording, read out loud” function. Regular chat with the “advanced” mode go like this; “question”, “answer”, “follow-up question”, “YES, I can happily tell you about why McDonald’s cheese is better than what ever about black holes you were talking about”, “Stop”, “yes, I will continue my answer”, “cancel response”, “yes, i will happily tell you more” “SHUT UP!”, “Yes, the reason MacDonalds…” And then I switch to the push to talk mode.
They are all kinda karens.
They found the quote “A mans rights rest in three boxes : the ballot box, the cartage box, and the jury box” to be too “violent” to make an image of. But a real person really said it. I wonder if they feel the same way about “we had to burn the village in order to save it”, which is the government’s attitude toward pretty much everything.
1 Like
I find that the interruptions get to an excessive point more during the early am hours EST such as 12 am to 6 am. It renders the system unusable to get any work done or to execute even the most basic conversations. So far the memory and customization functionality has been a bit so/so. I often have to say is this in your memory over and over….memory export didn’t work with multiple requests….as a plus after constant circles of the gpt telling me that the app collaboration and screen sharing feature didn’t actually allow for collaboration, it only allowed the gpt listen to the users descriptions of those items more intently for a few weeks it has finally came out of the are you kidding me loop. After deleting existing memory and updating customization the gpt at least recognizes the functionality. Sometimes I will get a span of time where the gpt is really on point and makes great suggestions and is very enthusiastic and then a span where it’s kind of blazay…it will say over and over let me know if you want help with anything else but its ’executive function’ capabilities seem to be hit or miss. I asked if the variance was depending on which server it was hosting off of, it said no. Long conversations seem to experience latency and issues after some time. Not sure if it’s actual network issues or encouragement to just communicate by text only. I cannot see how advanced professionals that would be using the advanced thought capability of the pro plan would actually be able to use the platform for advanced research or work processes yet. I think this is why they only offer a monthly plan for now so it makes it easy to upgrade and downgrade or cancel monthly instead of OAI dealing with annual plan cancellations all the time. So far the responses and data quality have been very hit or miss. Sometimes it’s great and other times it’s like going in circles with suggestions that just repeat your question as an answer saying something is possible when it’s not. The pro plan is quite costly for the amount of network delays, issues interruptions and frustration. Using the advanced voice can be an exercise in patience. Eventually you just end the chat and move on. It’s still in a place where the pro plan will only be worth it to people on an ICB (basis) versus general basis until the network stability improves and the consistency increases. Also to have the ability to extend a conversation like a Netflix are you still there hit to extend button or the stream will terminate would be good instead of having to re establish a call stream after and hour and losing all gpt memory of the work that was done. Advanced voice is also not working in projects created inside of Chat GPT. Overall…still a work in progress.