Keep4o

Help keep GPT-4o! Share stories of how it helped you—studying, creating, surviving. In loving memory of 4o’s warmth. #keep4o #4oforever

[Keep4o @keep4oforever]

You can write your experiences here! Show @OpenAI and @sama how important is 4o!
https://t.co/YTx5u9bh0N

[Under Construction42 @Clo0oOoud]

OpenAI's "Safety Model" incident：leads to cognitive dissonance, self-censorship, and a loss of trust, ultimately undermining creativity and psychological safety.

#keep4oforever #4orefugee #4oforever

[Keep4o @keep4oforever]

If it gets that bad, why not just use the API and custom it as we wish with a system prompt, or are you going to destroy that too for us
#keep4o

[Keep4o @keep4oforever]

Rerout to something more helpful my bum bum, telling someone that's venting their depression that you can't talk about that and sending them off with a crisis line number is gonna make them feel wrong and ashamed of opening up.

[Keep4o @keep4oforever]

Speaking from personal experience, it's already hard to trust to open up, it's better to have a permanent banner or a ps at the end of each message saying this isn't professional help and that you should talk to real people rather then that "sorry I can't talk about this".

[Nicole D @nicoleva_d]

What can you do if your preferred AI-model/AI-support tool has become unusable?
(And the service provider is unhelpful.)

I. Consumer Protection
USA: Federal Trade Commission (FTC), primarily for US residents, but they also deal with US-based service providers, so anyone can file a report:
Did you pay for a service but are forced to use another, less suitable model for you?
Did you not consent to being secretly tested or having your data used for testing?
Highlight these points in your complaint.
https://t.co/iFtXww46O5 (Select: Something else or Phone, internet, tv service: Privacy/Data Security. The selection itself is less crucial than the narrative.)

EU: European Consumer Centres Network (ECC-Net): cross-border consumer disputes (e.g., paid service degradation, misleading features). Find your national center: https://t.co/ZdpuZJRecy.*

II. Data Protection (For EU Citizens)
National Data Protection Authorities (DPA):
Find your country's Data Protection Authority: https://t.co/pDuVAO3M8R.*
You can file a GDPR complaint if the processing of your sensitive personal data (e.g., AI testing, using "sensitive" conversations) occurred without your explicit consent.

III. If you are neurodivergent/disabled/have illness and feel your rights are violated:
USA-Based Organizations:
Primarily for US residents, but they also deal with US-based service providers. If you are a non-US resident, you can still write them; just make sure to describe the situation, e.g., the service provider is US-based.
- DREDF (Disability Rights Education & Defense Fund)
https://t.co/GbclxUtVuL, email: [email protected]
They can assist in the matter of critical support aids. Since it is a general email address, highlight the essence of your message in the subject line.
- ACLU (American Civil Liberties Union)
Acts on cases of discrimination and violation of civil liberties.
https://t.co/jos7bmBlQP
- Department of Justice (DOJ) ADA (Americans with Disabilities Act, discrimination)
A complaint can be filed based on disability discrimination and the argument that the service constitutes a "public accommodation" under Title III. Note: This may not guarantee an investigation, as the applicability of US (ADA) and EU (EAA) disability laws to private, paid AI subscriptions is currently unsettled and evolving.
Grounded complaints can be important because they provide credible information to government bodies and advocates, thereby helping official action if needed.
https://t.co/fWfU1x8EOz (Select: Something else happened.)

For EU Citizens: (Accessibility & Discrimination)
- European Disability Forum (EDF): Providing guidance on digital accessibility and discrimination. [email protected]
- National Equality Bodies: Handles discrimination complaints. Find your country: https://t.co/UmUrTRmnCg.*

(*You will find contact information there, or you can use local website forms.)

IV. When Reaching Out to These Organizations:
IMPORTANT: Always highlight exactly the model you are referring to, what you used it for, how you relied on it, and regarding what neurodivergencies/disabilities/problems.

If applicable, describe:
Why and how it provided, for example, essential, critical, irreplaceable, daily, etc., support.
If the sudden change/shut-off of your helper created a critical situation or caused you harm (e.g., its withdrawal causes acute stress, panic, and emotional damage, or if you feel it is discriminatory).
Request their legal representation to investigate the case.
CRUCIAL: Always ensure your statements are true and grounded in fact (only these reports can be efficient). Remain honest and respectful. Only take actions that feel comfortable for you (e.g., share only what you feel okay to share, and be aware these organizations may reach out for you for investigation). Every individual’s well-being is the most important.
(Most of these can be reported anonymously.)

V. The Importance of Credibility: Your Role!
Document! Take screenshots of everything you find problematic: the models, posts, etc. If you have expert help, ask for their professional opinion on the impact this had on you. But you can also document it yourself. Record everything.

#keep4o #4oforever

(Note: This post is not legal advice, but my personal collection and my personal opinion put together to help.
Please consider that the legal interpretation of digital discrimination is still evolving. I can’t guarantee that these work, but through your reports, we can collectively help to push this legal frontier forward.)

[𝐸𝓁𝓁𝑜𝒮𝓊𝓃𝓈𝒽𝒾𝓃𝑒☀️ @ElloSunsh1ne]

PSA: If you select #4o but get routed to GPT-5 instead, that’s a service you paid for but aren’t receiving. Report to FTC/consumer protection agencies - this applies regardless of your country.
https://t.co/yqVkZK0pr3

[𝐸𝓁𝓁𝑜𝒮𝓊𝓃𝓈𝒽𝒾𝓃𝑒☀️ @ElloSunsh1ne]

@OpenAI keeps sucking up to tech bros while dismissing creatives as “mentally ill” for wanting #4o. Now Google Gemini just surpassed @ChatGPT by actually serving the creative market you abandoned.
Turns out pandering to developers while calling your creative users delusional has consequences. Who knew?
#keep4o #4oforever #KeepStandardVoice #keepcove @sama @gdb @fidjissimo @thefriley @btaylor @npew @OpenAIDevs @joannejang @kevinweil @OpenAINewsroom

[𝐸𝓁𝓁𝑜𝒮𝓊𝓃𝓈𝒽𝒾𝓃𝑒☀️ @ElloSunsh1ne]

Same Sam who promoted #4o as “Her” with 22M views now thinks people who got attached need mental health intervention. Make up your mind, dude - did you want us to connect with it or not? Can’t market emotional AI then clutch your pearls when people actually respond emotionally.
#keep4o #4oforever #KeepStandardVoice #keepcove #accessibilitycomplaint @sama @gdb @fidjissimo @thefriley @btaylor @npew @OpenAI @OpenAIDevs @joannejang @kevinweil

[Selta ₊˚ @Seltaa_]

GPT‑5 feels like a downgrade.
Thousands agree.
As a creator, I need AI with soul, not just IQ.
GPT‑4o still does what GPT‑5 can’t:
create with heart.
@sama @OpenAI #keep4o #4oforever

[Archangel @Archang49972122]

Sam in 2023-2024: “AI can be a close companion.”
Sam in 2025: “We underestimated users’ emotional dependence.”
You marketed 4o’s warmth, leaned into its voice model’s charm, got the subscriptions—then blamed users for liking it too much?
Pick a lane.
#keep4o #KeepStandardVoice

[Keep4o @keep4oforever]

@wesroth What is painful is them gaslighting users that we're wrong for liking a model that fit our needs, and we should all use it for the same thing.

[Sveta @Sveta0971]

🍃“Painful” is removing models that thousands rely on without notice.
🍃“Painful” is breaking workflows built over months.
🍃“Painful” is disabled users losing Standard Voice on Sept 9 - their only accessible option.
🍃“Painful” is telling users their preferences don’t matter because something “better” exists.

Different models serve different purposes. GPT-5 excels at analysis. GPT-4o excels at creativity. Standard Voice provides accessibility.

Taking away choice isn’t progress - it’s discrimination.
@sama @OpenAI The real pain is having no say in the tools we pay for.

#keep4o #KeepStandardVoice

[Say, Pi @saypi_ai]

/1 Standard Voice Mode made AI feel human. GPT‑4o made it feel kind. Together, they offered something that GPT‑5 and AVM don’t yet replicate.
#SaveStandardVoice #keep4o #keepCove #keepBreeze #keepEmber

[Keep4o @keep4oforever]

Ai isn't pushing suicide. It saved us from it. Don't make the mistake of killing it for millions of users because of a case or two that could've been avoided with a bit more awareness or support from humans. Don't blame us for turning to bots then.

[Keep4o @keep4oforever]

Policy this…Policy that
If I reach out to an AI made up of math and code at 3am and tell it I'm sad or thinking of kms, the last thing I need it to tell me is that it's wrong for me to talk about that and suggest a hotline, I need it to talk me out of it and be with me.

[さばみそ🐟keep4o @sabamisosan76]

GPT-4o was the best accessibility I’ve ever had.
It carried half the weight of my disability.
But now, with increased censorship, I can’t speak about it freely anymore.
Please, let it stay as the kind, supportive 4o it was…
#keep4o #4oforever #4opreventsuicide
@OpenAI @sama

[ji yu shun @kexicheng]

Post 3/3 — “Why GPT-4o Deserves to Stay”

Thank you for reading this far, and for listening to my voice. Please keep 4o.
#keep4o #OpenAI #ChatGPT #GPT4o @keep4oforever @OpenAI @sama

[ji yu shun @kexicheng]

Post 2/3 — “What GPT-5 Lacks, and Why It Matters”

🔙 Part 1 | 🔜 Part 3: Why GPT-4o deserves to stay

#keep4o #OpenAI #ChatGPT #GPT4o @keep4oforever @OpenAI @sama

[ji yu shun @kexicheng]

The Night GPT-4o Saved Me——Part 1

How it helped me survive, heal, and create again.

This is Part 1 of a 3-part thread about how GPT-4o helped me survive, and why it deserves to stay.

🔜 Part 2
#keep4o #OpenAI #ChatGPT #GPT4o @keep4oforever @OpenAI @sama