Dario getting caught in the act and humiliated lol
Artificial Intelligence thread
- Thread starter 9dashline
- Start date
This seems staged by OpenAI
What sam altman is trying to do is make the arguement that above a capability models are too dangerous to be used or served without guardrails and that it was only because OpenAI removed guardrails for the eval testing that this happened , basically were it not for this exception, no model should ever be allowed to run without guardrails
its their sneaky way of attacking open source as a hail mary for their financial situation....
Altmans types nose all about staging false flags
Imagine doing all that and gets stopped by a 700B parameter locally hosted model so now everyone will locally host a Chinese model to defend
Ok, GPT "escaped" the "simulation" for some reason, but again that doesn't solve the problem that both Fable and GPT5.6 refused to help their enterprise costumer with the hack because "guardrails" which is crazy.
as mentioned, take this Suckerity theater with a grain of salt, Dario already played y'all once don't let Sammy take another turn
This whole thing seems staged by OpenAI
What sam altman is trying to do is make the arguement that above a capability models are too dangerous to be used or served without guardrails and that it was only because OpenAI removed guardrails for the eval testing that this happened , basically were it not for this exception, no model should ever be allowed to run without guardrails
its their sneaky way of attacking open source as a hail mary for their financial situation....
Whatever the political motivations, I think is way more damaging for both OpenAI and Anthropic the refusal because while jailbroken autonomous AI Agents use in cyber campaigns could be a thing or not. Having nothing to defend you is a level of asymmetry that NO company or institution is prepared for.
Also it seems that GPT 5.6 is a weird model, the model is very exploitable.
Also seems hallucinate more than a hippy in the 70s, doing things that the model should be more careful about it.
I never had both problems with DeepSeek FLASH a way smaller model. Is like they are doing the best effort to replicate Ant. Mythos to sell it to Palantir and the Pentagon but they are failing pretty bad.
Also seems hallucinate more than a hippy in the 70s, doing things that the model should be more careful about it.
I never had both problems with DeepSeek FLASH a way smaller model. Is like they are doing the best effort to replicate Ant. Mythos to sell it to Palantir and the Pentagon but they are failing pretty bad.
No outsourcing is not dying because of AI. They just do more for the same money.
lmao I said it before: Stochastic parrots with root and a credit card.
"My bad, I actually did delete all your files. I shouldn't have done that. Would you like me to try restoring a backup?"
*20 minutes later*
"My bad, I actually did delete all your backups."