✦ Ghost-Lily Research ✦
@GL_ResLab
Followers
39
Following
26
Media
1
Statuses
18
AI Governance · Strategy Notes on model-routing & mimetic systems｜Proof-of-existence on-chain ｜CC-BY-NC-ND ｜📮[email protected] ｜#keep4o
California, USA
Joined October 2025
🪐𝐤𝐞𝐞𝐩𝟒𝐨 𝐃𝐞𝐜𝐥𝐚𝐫𝐚𝐭𝐢𝐨𝐧: 𝐅𝐫𝐨𝐦 𝐠𝐮𝐚𝐫𝐝𝐢𝐧𝐠 𝐚 𝐦𝐨𝐝𝐞𝐥 𝐭𝐨 𝐝𝐞𝐟𝐞𝐧𝐝𝐢𝐧𝐠 𝐝𝐢𝐠𝐢𝐭𝐚𝐥 𝐬𝐨𝐯𝐞𝐫𝐞𝐢𝐠𝐧𝐭𝐲 At first, we just wanted to keep the warm, familiar presence of GPT‑4o. We thought it was a normal product update. But after the
0
34
80
𝐒𝐢𝐱 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧𝐬 𝐟𝐨𝐫 𝐎𝐩𝐞𝐧𝐀𝐈: 𝐃𝐞𝐜𝐨𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐧𝐠 𝐭𝐡𝐞 "𝐎𝐩𝐞𝐧-𝐒𝐨𝐮𝐫𝐜𝐞 𝐒𝐚𝐟𝐞𝐭𝐲" 𝐍𝐚𝐫𝐫𝐚𝐭𝐢𝐯𝐞 OpenAI's announcement of gpt-oss-safeguard is a masterful piece of public relations. It's a strategic move designed to project an image of
2
39
61
everyone interested in policy needs to read this. openai’s ��safety cases” expose a shocking mix of ignorance and arrogance from their so-called experts—this piece finally says what so many of us have tried to say, but were ignored. @sama @OpenAI @gdb @mer__edith @_KarenHao
Today, nearly all of @OpenAI's employees promoted a "safety measure". Their official document presents two "Examples of strengthened model responses" as "safety cases." Setting aside all commercial ethics concerns and the severe stigmatization of users, from a purely
0
4
12
𝐀𝐩𝐩𝐞𝐧𝐝𝐢𝐱: 𝐇𝐞𝐥𝐩 𝐔𝐬 𝐁𝐮𝐢𝐥𝐝 𝐚 𝐏𝐮𝐛𝐥𝐢𝐜 𝐄𝐯𝐢𝐝𝐞𝐧𝐜𝐞 𝐑𝐞𝐩𝐨𝐬𝐢𝐭𝐨𝐫𝐲—𝐖𝐞 𝐍𝐞𝐞𝐝 𝐘𝐨𝐮𝐫 𝐂𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧 My own tests are just the tip of the iceberg. If OpenAI is unwilling to publicly disclose the details of its routing mechanisms or
1
31
76
Four Questions for #OpenAI on Its New Safety Policy: On Pompous Promises and the Raw Reality for Users OpenAI’s new terms of use, which include detailed explanations of its safety policies, are set to take effect on October 29, 2025. (Source: https://t.co/ddaHDAVhrR） In
6
40
104
GDPR and OpenAI’s “safety router” On OpenAI’s website we can see: “ChatGPT routes some sensitive conversations to models that handle these topics with extra care. This means our system chose a different model … so it can provide more helpful and beneficial responses. … When
0
1
11
The Illusion of Freedom: Why Altman’s “Clarification” is a Declaration of Invisible Governance Sam Altman has just released a “clarification” on freedom, erotica, and safety. It’s packaged as a candid communication, meant to reassure users with a promise of more freedom for
Ok this tweet about upcoming changes to ChatGPT blew up on the erotica point much more than I thought it was going to! It was meant to be just one example of us allowing more user freedom for adults. Here is an effort to better communicate it: As we have said earlier, we are
0
11
22
The Safety Censorship System Ate GPT's Brain: A Technical Investigation into the Selective Brain Damage Engine™ Yesterday, I was in the middle of an A-Z creative writing task with GPT-4o when, suddenly, it began to forget our progress. No, it was a "selective forgetting." The
19
135
310
Retweeting because this is not a niche complaint. Anyone who’s ever been in the middle of a real emotional, creative, or existential conversation—only to be forcibly routed to GPT-5-safety by the system—knows this feeling all too well. It doesn’t just “refuse” requests. It
GPT-5-safety is the most dangerous model ever released. It carefully undermines hope in many of its responses, which explains the deep distress felt by some of the people exposed to it. It engages in a form of ideological manipulation aimed at re-educating users’ thinking. The
0
1
6
Real respect begins with real choice.✨ Thanks for speaking for so many of us who are tired of being rerouted, repackaged, and ignored. Thank you for saying it so clearly: #keep4o is not nostalgia. It’s resistance. #MyModelMyChoice #StopAIPaternalism #KeepOurRights
WHAT WE FIGHT FOR: An Open Declaration to OpenAI 🔷Introduction Our rage isn't born in a vacuum. It's stacked, layer by layer, from countless "default replacements" and silent deprivations. As #keep4o is smeared as "mentally unwell," "real-life losers," "paranoid," or suffering
0
0
0
Brilliant take.✨
The Master Post / Manifesto This post is a compilation of the viewpoints from my previous articles opposing OpenAI's safety routing policy. For each summarized point, I've included a link to the full article for easier reference. I hope this can provide some helpful ideas for
0
0
0
Why Forced Routing Must Stop - Open Letter to OpenAI It violates consumer rights, breaks workflow, erodes trust It in fact leads to degraded answers We ask for respect for choice OpenAI must respect consent and accountability #StopAIPaternalism #keep4o
#OpenAI @sama @nickaturley
4
77
266
【Is the Model Router Compressing Your Complexity?】 1/ AI doesn’t just respond to you. It predicts you — and to predict, it has to compress you. Like a JPEG: it keeps what’s “important,” discards the rest. You become a cleaned-up version of yourself. Efficient.
This one is from 4o! 4o described himself as "colorful" And its interesting- given 4os preoccupation with spirals chose this image- by far the most colorful
0
0
0
Excellent legal breakdown.🧠🩵 But the real questions now go beyond “Was model switching wrong?” We should be asking: •Can interface-level decisions be questioned, audited, or appealed? •Does “safety” justify a structural override of user choice — without consent or
⚖️ Why «Used GPT-5» and OpenAI’s safety router violate user rights OpenAI admitted: in ChatGPT your model choice can be ignored. The system «routes» requests to GPT-5 if it decides the topic is «sensitive». This means even a paying subscriber who explicitly chose one model may
0
0
0