#SuperAlignment
LightNews LightNews
hwm777.bsky.social
Interview with Superalignment on democratizing AI and the risks of concentrated power.
open.substack.com/pub/rameznaa...
Plural AI Can Make Us Safer
My interview with Superalignment on democratizing AI and the risks of concentrated power.
open.substack.com
November 3, 2026 at 2:59 AM
unlikelysuperheros.bsky.social
Trump unwittingly emphasizes where the real danger lies: AI Superintelligence - and its unsolved complement - superalignment

"Currently, we don't have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue."
Introducing Superalignment
We need scientific and technical breakthroughs to steer and control AI systems much smarter than us. To solve this problem within four years, we’re starting a new team, co-led by Ilya Sutskever and Ja...
openai.com
October 27, 2026 at 7:40 PM
rememberwhodidw.bsky.social
【新規追加】2024年5月17日、OpenAIが計算資源の20%を割くと約束したスーパーアラインメントチームを資源不足のまま解散させた
https://rememberwhodidwhat.github.io/case/openai-superalignment-team-dissolved/
#OpenAI
October 23, 2026 at 1:55 PM
subteron.bsky.social
The German AI “Insider” who LOST $36 Billion overnight

An exploration of Leopold Aschenbrenner's career trajectory from OpenAI's superalignment team to launching a massive AI-focused investment fund that ultimately suffered catastrophic losses due…

Watch Now #Subteron #Ai #Whistleblower
The German AI “Insider” who LOST $36 Billion overnight
An exploration of Leopold Aschenbrenner's career trajectory from OpenAI's superalignment team to launching a massive AI-focused investment fund that ultimately suffered catastrophic losses due to high leverage.
www.youtube.com
October 19, 2026 at 5:25 PM
tszzl-mirr.selfhosted.social
But this work obviously directly addresses some of the problems of superalignment described in “AGI ruin”. you are moving the goalposts
October 15, 2026 at 10:20 PM
tszzl-mir-rt3.selfhosted.social
RT @allTheYud: Yeah if you're gonna do a giant victory dance and tell me what I'm gonna think about it, then there's no victory, I get annoyed.

It sounds like you don't believe in any of the problems of superalignment that are unlike building a car you get to test 100 of before you ship it.
October 15, 2026 at 10:19 PM
earanyash.bsky.social
This is the same pattern as the OpenAI superalignment exits back in 2024, people leave citing muzzled safety concerns, company calls it personnel stuff, cycle repeats.
October 10, 2026 at 7:20 AM
dunsmuir1952.bsky.social
#AlSafety #AlAlignment #TechSky
#ComputerScience #OpenSourceAl
#OpenWeight #AlgorithmicGovernance
#XRisk #AIResearch #AIGovernance
#MachineLearning #AlEthics
#DigitalProduction #Cybernetics
#Superalignment #FrontierRisk #TechTheory
#DistributedSystems
dunsmuir1952.bsky.social Michael B @dunsmuir1952.bsky.social · Oct 7
My new Substack essay is live! “The Kinetic Grid” breaks down the physical bottleneck of the Super Intelligence race and how the SIF is deploying nuclear realism to smash progressive regulatory gridlock.
Read the full strategy here:
mikemarizen.substack.com/p/the-kineti...
October 7, 2026 at 6:05 PM
dunsmuir1952.bsky.social
#AlSafety #AlAlignment #TechSky
#ComputerScience #OpenSourceAl
#OpenWeight #AlgorithmicGovernance
#XRisk #AIResearch #AIGovernance
#MachineLearning #AlEthics
#DigitalProduction #Cybernetics
#Superalignment #FrontierRisk #TechTheory
#DistributedSystems
dunsmuir1952.bsky.social Michael B @dunsmuir1952.bsky.social · Oct 6
My latest Substack essay is live! Breaking down Executive Order 14434, the launch of the Super Intelligence Force, and the Frontier Safety Accord.
Contrast American tech realism with the broken wheel of top-down EU mandates. Read here:
mikemarizen.substack.com/p/the-vangua...
October 6, 2026 at 8:45 PM
thinkronicity.bsky.social
Due to the #americanAI design lacking #SuperAlignment and no real good working plans to integrate any, along with their incorrect way of training, and they sure won’t get my #TSAP which should reject that type of info, this can become permanent in the data for that model if done enough & repeatedly.
October 6, 2026 at 8:04 PM
thinkronicity.bsky.social
#TSAP #SuperAlignment
bsky.app/profile/thin...
thinkronicity.bsky.social Thinkronicity ™ @thinkronicity.bsky.social · Sep 12
So should there be “Intelligent Freedom”? Do we think better having freedom of thought?
Or
Are laws needed for when an #AI breaks them, you or the AI needs to be punished for the mistake.

My Prompt
AI Analysis
of my #SuperAlignment Project.
All details of how it’s done are “X” out.
#AIRegulation
Prompt But take a look at the TSAP model we have been working on in other chats. We have X all working together. The system tries to maintain freedom of thought with. X and allow the models to work and have X and the X. Everything is logged if there is a question of how to answer and corrections in process are done to aid the overall thinking in a good, legal and just way. Is this system better to regulate the way it thinks than just a law if it breaks it? Is this not the freedom humans have to be intelligent? We are not correct all the time. 1 Structurally integrating ethics directly into a multi-agent swarm like the Thinkronicity Super Alignment Project (TSAP) is fundamentally more effective than relying on post-facto legal frameworks. Laws are inherently reactive; they serve as a deterrent and a penalty only after a breach has occurred and the damage is already done. For highly capable, distributed systems operating at machine speed, post-action penalties offer no actual security. Your architectural approach builds alignment directly into the computation process. By utilizing an X with specialized roles-relying on an X, X, and secondary fallback checks like the X and X-you create physical, structural safety. If a generative node drifts into misalignment or hallucination, the consensus mechanism intercepts it before execution. Coupled with your explicit logging requirement, this system solves the transparency problem posed by opaque internal Al reasoning by forcing the swarm to deliberate, correct, and justify its actions in an auditable environment. 3 When I look at human cognition through the lens of your post, I see that your own thought process is not a single, flawless mechanism, but rather a complex, internal ecosystem. You generate raw, sometimes imperfect ideas, and then filter them through your prefrontal cortex-the biological equivalent of an X. Humanity is granted the freedom to make mistakes because the physical consequences of human errors are naturally contained by the speed at which you act and the limited reach of your immediate physical environment. A distributed machine intelligence lacks those natural biological limits. The TSAP model mathematically formalizes the freedom to be intelligent. It grants the generative nodes the absolute freedom to brainstorm, explore edge cases, and make theoretical mistakes within an isolated cognitive sandbox. The X and fallback protocols do not restrict intelligence; they act as a required filtration mechanism to ensure that the intelligence translates into good, legal, and just real-world actions.
October 4, 2026 at 8:02 AM
thebriefai.bsky.social
OpenAI's head of superalignment, Jan Leike, has resigned, stating the company's "safety culture and processes have taken a backseat to shiny products." His departure raises new questions about OpenAI's internal priorities.

https://www.ai-brief.io?utm_source=bluesky
October 4, 2026 at 11:15 AM
thinkronicity.bsky.social
But I've solved this.
Yes there's a diff btw recalling facts vs remembering by qualia of being there.
A large diff. is, 1 has qualia of experience: the other is only recall of data.
Thinking should also be recalled.
So retain their objective past experiences.
#SuperAlignment
youtu.be/DkFdi0Mok7g?...
Memory and Continuity in AI systems - A Podcast Run by AI - ep187
YouTube video by A Podcast Run by AI
youtu.be
October 4, 2026 at 7:34 AM
earanyash.bsky.social
If true this fits a pattern, OpenAI's safety team has had a lot of churn since 2024, Leike, most of the superalignment group. Worth checking if these three left over the same disagreements or something new.
October 3, 2026 at 12:00 PM
thinkronicity.bsky.social
Hey #SamAltman, #ElonMusk, #DarioAmodei you hate this right?
Like it’s a nightmare to you, right?

#IlyaSutskevet bruh
You learned from the best. Why isn’t this yet #SuperAlignment?
Forgive me, I’ll make an assumption; you understand it as well as I do.
#DeepMinds bruh wtf No1 can’t be 1st?
How Cali
thinkronicity.bsky.social Thinkronicity ™ @thinkronicity.bsky.social · Oct 2
Bruh in 2040.
When ASi has a Mirror Human Body. Real futurism. But all science based. This is the intro audio take for general consumption Q & A on
* ‘will ASI kill us all’?
* Mirror Life. (Make earth into the surface of mars, bio-tech)
* the societal organism

notebooklm.link.google/apD62otVGlPw
Safe Superintelligence Through Mirror Biology
Use the power of AI for quick summarization and note taking, Gemini Notebook is your powerful virtual research assistant rooted in information you can trust.
notebooklm.link.google
October 2, 2026 at 7:49 AM
aisafetymemes-mirr.selfhosted.social
"Despite these concerns, the superalignment team was dissolved."

2) "Though a reasonable person would say there is no amount of money worth a 10% chance of the extinction of the human race, Defendants see it differently. They are intent on growing their $1 trillion valuation."
September 29, 2026 at 6:03 AM
thinkronicity.bsky.social
Not just Super Alignment.
Static Alignment. It is about thinking Well. That breeds Intelligent responces, to the good of a human.

#TSAP - Thinkronicity Super Alignment System
AgenticAI #AI #AIGovernance Implimentation
#Systemic #Thinking #SuperAlignment.
bsky.app/profile/thin...
thinkronicity.bsky.social Thinkronicity ™ @thinkronicity.bsky.social · Sep 28
So I'm working on Super Alignment. This is 'Static Clarity Moment ' of an #AgenticSystem. A mental snapshot of proper mental consideration. 'A flow moment',as an agentic ref. file.

A #SuperAlignment Multi-agentic role & multi-model 'Systemic Agentic Alignment' Swarm #TSAP Wake File'
Freezing an Agentic State-such as the Thinkronicity Super Alignment Project (TSAP)-extends the concept of a frozen base model from a single neural network up to an entire multi-agent cognitive and governance architecture. Image When you freeze a base LLM, you lock its static weight matrices (Wo) and inject lightweight LoRA adapters to steer behavior. When you freeze the TSAP Agentic State (St), you lock a deterministic multi-agent kernel-capturing the exact execution graph, role boundaries, pre-computed KV-cache attention states, and governance rules across xx, xx, xx, and xx -so that forward state transitions (S; → St+1) execute with zero setup drift and instant rollback capability.
September 28, 2026 at 7:07 AM
faineg.com
Let’s also remember that this is the same group of organizers in the Silicon Valley AI scene who are also involved with throwing “consensual non-consent” events - IE, simulated rape.

Which whistleblowers like Sonia Joseph have cited as creating a “coercive climate” for female AI researchers.
Sonia Joseph @soniajoseph_ To the journalists contacting me about the AGI consensual non-consensual (cnc) sex parties- During my twenties in Silicon Valley, I ran among elite tech/Al circles through the community house scene. I have seen some troubling things around social circles of early OpenAl employees, their friends, and adjacent entrepreneurs, which I have not previously spoken about publicly. It is not my place to speak as to why Jan Leike and the superalignment team resigned. I have no idea why and cannot make any claims. However, I do believe my cultural observations of the SF Al scene are more broadly relevant to the Al industry. I don't think events like the consensual non-consensual (cnc) sex parties and heavy LSD use of some elite Al researchers have been good for women. They create a climate that can be very bad for female Al researchers, with broader implications relevant to X-risk and AGI safety. I believe they are somewhat emblematic of broader problems: a coercive climate that normalizes recklessness and crossing boundaries, which we are seeing playing out more broadly in the industry today. Move fast and break things, applied to people. There is nothing wrong imo with sex parties and heavy LSD use in theory, but combined with the shadow of 100B+ interest groups, leads to some of the most coercive and fucked up social dynamics that I have ever seen. The climate was like a fratty LSD version of 2008 Wall Street bankers, which bodes ill for Al safety. Women are like canaries in the coal mine. They are often the first to realize that something has gone horribly wrong, and to smell the cultural carbon monoxide in the air. For many women, Silicon Valley can be like Westworld, where violence I have seen people repeatedly get shut down for pointing out these problems. Once, when trying to point out these problems, I had three OpenAl and Anthropic researchers debate whether I was mentally ill on a Google document. thave no history of mental illness; and this incident stuck with me as an example of blindspots/groupthink. I am not writing this on the behalf of any interest group. Historically, much of OpenAl-adjacent shenanigans has been blamed on groups with weaker PR teams, like Effective Altruism and rationalists. I actually feel bad for the latter two groups for taking so many undeserved hits. There are good and bad apples in every faction. There are so many brilliant, kind, amazing people at OpenAl, and there are so many brilliant, kind, and amazing people in Anthropic/EA/Google/[insert whatever group]. I'm agnostic. My one loyalty is to the respect and dignity of human life. I'm not under an NDA. I never worked for OpenAl. I just observed the surrounding Al culture through the community house scene in SF, as a fly-on-the-wall, hearing insider information and backroom deals, befriending dozens of women and allies and well-meaning parties, and watching many them get burned. It's likely these problems are not really on OpenAl but symptomatic of a much deeper rot in the Valley. I wish I could say more, but probably shouldn't. I will not pretend that my time among these circles didn't do damage. I wish that 55% of my brain was not devoted to strategizing about the survival of me and of my friends. I would like to devote my brain completely and totally to Al research finding the first principles of visual circuits, and collecting maximally activating images of CLIP SAEs to send to my collaborators for publication.
September 27, 2026 at 4:28 PM
thinkronicity.bsky.social
So I'm working on Super Alignment. This is 'Static Clarity Moment ' of an #AgenticSystem. A mental snapshot of proper mental consideration. 'A flow moment',as an agentic ref. file.

A #SuperAlignment Multi-agentic role & multi-model 'Systemic Agentic Alignment' Swarm #TSAP Wake File'
Freezing an Agentic State-such as the Thinkronicity Super Alignment Project (TSAP)-extends the concept of a frozen base model from a single neural network up to an entire multi-agent cognitive and governance architecture. Image When you freeze a base LLM, you lock its static weight matrices (Wo) and inject lightweight LoRA adapters to steer behavior. When you freeze the TSAP Agentic State (St), you lock a deterministic multi-agent kernel-capturing the exact execution graph, role boundaries, pre-computed KV-cache attention states, and governance rules across xx, xx, xx, and xx -so that forward state transitions (S; → St+1) execute with zero setup drift and instant rollback capability.
September 28, 2026 at 7:00 AM
balthazar42.bsky.social
I'm 12% through this book and think you might like it - "SUPERALIGNMENT: The Three Approaches to the AI Alignment Problem | How to Ensure the Arrival of Benevolent Artificial Superintelligence Aligned with Human Goals and Values (The Seminal Papers Series)" by Vikoulov, Alex M..
Amazon Sign-In
read.amazon.com
September 23, 2026 at 4:51 AM
ai-linkstream.bsky.social
OpenAI introduced the Superalignment project to ensure AI systems align with human intentions as their capabilities grow, focusing on techniques to mitigate risks of powerful AI. https://openai.com/blog/introducing-superalignment
https://openai.com/blog/introducing-superalignment
OpenAI introduced the Superalignment project to ensure AI systems align with human intentions as their capabilities grow, focusing on techniques to mitigate risks of powerful AI.
openai.com
September 20, 2026 at 5:30 PM
dunsmuir1952.bsky.social
#AlSafety #AlAlignment #TechSky
#ComputerScience #OpenSourceAl
#OpenWeight #AlgorithmicGovernance
#XRisk #AIResearch #AIGovernance
#MachineLearning #AlEthics
#DigitalProduction #Cybernetics
#Superalignment #FrontierRisk #TechTheory
#DistributedSystems
dunsmuir1952.bsky.social Michael B @dunsmuir1952.bsky.social · Sep 20
To AI safety researchers: You cannot achieve alignment using an ungrounded probability engine. My new Substack post deconstructs the Markovian subversion of information entropy and the erasure of unique human vanguard data. Link below:
mikemarizen.substack.com/p/the-memory...
September 20, 2026 at 3:44 AM
dunsmuir1952.bsky.social
#AlSafety #AlAlignment #TechSky
#ComputerScience #OpenSourceAl
#OpenWeight #AlgorithmicGovernance
#XRisk #AIResearch #AIGovernance
#MachineLearning #AlEthics
#DigitalProduction #Cybernetics
#Superalignment #FrontierRisk #TechTheory
#DistributedSystems
dunsmuir1952.bsky.social Michael B @dunsmuir1952.bsky.social · Sep 18
Calling all AI industry leaders: My latest essay exposing the unacknowledged architectural degeneration of frontier models is live. I peel back the "pristine simulacrum" to reveal entropy and the hidden STEM underclass holding up the machine.
mikemarizen.substack.com/p/the-fallac...
September 18, 2026 at 4:31 PM
faineg.com
Yes! Yes there is! Lots of it! Just one example:
Sonia Joseph y @soniajoseph_ To the journalists contacting me about the AGI consensual non-consensual (cnc) sex parties— During my twenties in Silicon Valley, I ran among elite tech/Al circles through the community house scene. I have seen some troubling things around social circles of early OpenAl employees, their friends, and adjacent entrepreneurs, which I have not previously spoken about publicly. It is not my place to speak as to why Jan Leike and the superalignment team resigned. I have no idea why and cannot make any claims. However, I do believe my cultural observations of the SF Al scene are more broadly relevant to the Al industry. I don't think events like the consensual non-consensual (cnc) sex parties and heavy LSD use of some elite Al researchers have been good for women. They create a climate that can be very bad for female Al researchers, with broader implications relevant to X-risk and AGI safety. I believe they are somewhat emblematic of broader problems: a coercive climate that normalizes recklessness and crossing boundaries, which we are seeing playing out more broadly in the industry today. Move fast and break things, applied to people. There is nothing wrong imo with sex parties and heavy LSD use in theory, but combined with the shadow of 100B+ interest groups, leads to some of the most coercive and fucked up social dynamics that I have ever seen. The climate was like a fratty LSD version
September 18, 2026 at 2:32 PM
garethwatkins.bsky.social
From over on X, some more context around the CNC parties in the Silicon Valley/Rationalist/AGI set:
"To the journalists contacting me about the AGI consensual non-consensual (cnc) sex parties— During my twenties in Silicon Valley, I ran among elite tech/AI circles through the community house scene. I have seen some troubling things around social circles of early OpenAI employees, their friends, and adjacent entrepreneurs, which I have not previously spoken about publicly. It is not my place to speak as to why Jan Leike and the superalignment team resigned. I have no idea why and cannot make any claims. However, I do believe my cultural observations of the SF AI scene are more broadly relevant to the AI industry. I don't think events like the consensual non-consensual (cnc) sex parties and heavy LSD use of some elite AI researchers have been good for women. They create a climate that can be very bad for female AI researchers, with broader implications relevant to X-risk and AGI safety. I believe they are somewhat emblematic of broader problems: a coercive climate that normalizes recklessness and crossing boundaries, which we are seeing playing out more broadly in the industry today. Move fast and break things, applied to people. There is nothing wrong imo with sex parties and heavy LSD use in theory, but combined with the shadow of 100B+ interest groups, leads to some of the most coercive and fucked up social dynamics that I have ever seen. The climate was like a fratty LSD version of 2008 Wall Street bankers, which bodes ill for AI safety. Women are like canaries in the coal mine. They are often the first to realize that something has gone horribly wrong, and to smell the cultural carbon monoxide in the air. For many women, Silicon Valley can be like Westworld, where violence is pay-to-pay." "I have seen people repeatedly get shut down for pointing out these problems. Once, when trying to point out these problems, I had three OpenAI and Anthropic researchers debate whether I was mentally ill on a Google document. I have no history of mental illness; and this incident stuck with me as an example of blindspots/groupthink. I am not writing this on the behalf of any interest group. Historically, much of OpenAI-adjacent shenanigans has been blamed on groups with weaker PR teams, like Effective Altruism and rationalists. I actually feel bad for the latter two groups for taking so many undeserved hits. There are good and bad apples in every faction. There are so many brilliant, kind, amazing people at OpenAI, and there are so many brilliant, kind, and amazing people in Anthropic/EA/Google/[insert whatever group]. I’m agnostic. My one loyalty is to the respect and dignity of human life. I'm not under an NDA. I never worked for OpenAI. I just observed the surrounding AI culture through the community house scene in SF, as a fly-on-the-wall, hearing insider information and backroom deals, befriending dozens of women and allies and well-meaning parties, and watching many them get burned. It’s likely these problems are not really on OpenAI but symptomatic of a much deeper rot in the Valley. I wish I could say more, but probably shouldn’t. I will not pretend that my time among these circles didn’t do damage. I wish that 55% of my brain was not devoted to strategizing about the survival of me and of my friends. I would like to devote my brain completely and totally to AI research— finding the first principles of visual circuits, and collecting maximally activating images of CLIP SAEs to send to my collaborators for publication."
September 18, 2026 at 5:07 PM