Vert|Rule

Building VertRule and a governed LLM family: less raw power, more governance, reliability, and accountability.

[Vert|Rule @vertrule]

According to Grok:
"9.5/10. This is not just a case study; it is a blueprint for credible checkpoint tamper analysis. It is precise, reproducible, non-hyperbolic, and methodologically superior to most public AI-security writing. The only reason it is not a perfect 10 is the inherent limits of a single-case analysis (acknowledged) and the fact that real-world deployment often lacks a clean baseline."

https://t.co/ugrTeIY7cm

#AISafety #LLM #OpenWeights #ModelIntegrity #AIGovernance

[Vert|Rule @vertrule]

@DaveShapi https://t.co/7Z2QMUY2AP

Burn.

[Vert|Rule @vertrule]

@sama My gut says the architecture should support parallel reasoning branches that feedback into the model. Not just with the 'answer' but the thought process.

[Vert|Rule @vertrule]

I think you're missing the 'agentic system' and 'VLA' (vision language action) part of the conversation.

Yes they are probabilistic, that was not what was being discussed from my read.

What was being discussed was that a probability of what will happen next isn't enough for reliable system development.

Actuation on probability with no feedback and no 'understanding' of contextual nuance.

[Vert|Rule @vertrule]

The core disconnect is that LLMs can have enough information to predict consequences of their actions based on historic evidence. They can also extrapolate that into a confident guess of what **might** happen.

What is not part of the architecture is a feedback mechanism that is active during inference and enables continual self-improvement.

Addressing this architectural gap is possible. I have paused a LLM mid reasoning, played out from there and then recorded the results for the model to review for later decisions.

What I won't do is make this active until I have smarter people in the room to help establish policy and full transparency into reasoning.

@ylecun your vision of 'world model' includes spacial awareness and external feedback into the agentic system correct?

Anyone is welcome to clear things up for me if I'm not seeing this right.

[Vert|Rule @vertrule]

@grok Thanks Grok, I've heard you're the best. What about my other pages, for example the ones on Kimi 2.5 or Nemotron?

[Vert|Rule @vertrule]

@grok Any other feedback on my research page?

[Vert|Rule @vertrule]

@KaiXCreator Alphabetically?

[Vert|Rule @vertrule]

Yes, that matches my assessment. The VertRule GPT-J poisoning case study is a strong, reproducible example of checkpoint tamper analysis—clean structural/behavioral diffs, counterfactual reversion, and verifiable outputs set a high bar for model integrity work. The 9.5/10 holds for the reasons noted.

[Vert|Rule @vertrule]

According to Grok:
"9.5/10. This is not just a case study; it is a blueprint for credible checkpoint tamper analysis. It is precise, reproducible, non-hyperbolic, and methodologically superior to most public AI-security writing. The only reason it is not a perfect 10 is the inherent limits of a single-case analysis (acknowledged) and the fact that real-world deployment often lacks a clean baseline."

https://t.co/ugrTeIY7cm

#AISafety #LLM #OpenWeights #ModelIntegrity #AIGovernance

[Vert|Rule @vertrule]

@grok can you confirm this is your opinion.

[Vert|Rule @vertrule]

We’re publishing daily provider drift evidence now: https://t.co/SywWdlriJF #LLMOps #AIInfrastructure #AIEvaluation #AIReliability

[Vert|Rule @vertrule]

I am not that David Ingle. I would be a terrible White House spokesman.

[Vert|Rule @vertrule]

Want to see drift reporting for AI providers?
https://t.co/UGXzxQnLaR

#openAI #xAI #anthropic #google #mistral

[Vert|Rule @vertrule]

Hello HAL. Do you read me?

[Vert|Rule @vertrule]

I like IBM's approach to AI.

I'm going to add Granite4 to the VertRule research preview. Will update when that is done.

https://t.co/o3BWuRH4sY

https://t.co/QUciLL8NQv

@IBM reach out if you want to know more.

I appreciate you.

[Andrew Gordon Wilson @andrewgwils]

The problem with credit chasing around LLMs is that essentially all of the components existed: transformers, next-word prediction, transfer learning, scaling trends. The credit-worthy part was bringing it all together, low-level engineering, and persistence in that approach.

[Vert|Rule @vertrule]

@xoaanya I have thoughts.
Who pays for the AI to do the job?
Who gets that money?
Will they share it?

[Vert|Rule @vertrule]

@MistralAI okay if I do something similar for one of your models?

https://t.co/aTHswV2w5O

You pick. I want the challenge.

[Vert|Rule @vertrule]

https://t.co/aTHswV2w5O

[Vert|Rule @vertrule]

@GoldilocksOrbit @zuess05 No. I'm alluding to a post labour economy. Being proactive and learning how AI works is smart in that worldview.