My conspiracy about major network infrastructure going down is not a cyberattack
My bet is that the US government told OpenAI and Anthropic to have their agents target weaknesses within all these major networks. Then make them all patch up these weak points
Preemptive defense and testing against future real foreign cyberattacks
Keone Hon says a swap on Monad finalizes in 600 milliseconds. On Ethereum, finality is 12 minutes.
But speed isn't his first priority.
"The single most important thing is reliability. Like, the blockchain cannot have an outage."
No hacks. No faults. The state always has to be correct. Then you push performance.
"By comparison, when you do something on Ethereum right now, it's like 12 seconds, and then the finality is actually 12 minutes."
@keoneHD@monad
@GoingBallistic5@3blue1brown I remember watching this when it came out freshman year of HS. His animations and explanation style was way better and more intuitive than any teacher I’ve had
Wall Street execs and NSA have told Congress that outages were caused by patching vulnerabilities identified by AI models, per Washington Sun
AI did not cause outages but found so many issues that the rush to fix them before an attack overwhelmed systems
Another true conspiracy about Jensen Huang:
After the Hugging Face hack by OpenAI, the US government realized that using quality LLMs + a ton of compute, almost anything could be hacked.
The US government can’t stop Chinese LLM development, but they could limit chip exports to restrict high level compute access.
Jensen was backdooring Nvidia chips to China via his mistress and the government had to remove him for national security purposes.
The recent outages at Microsoft, FAA and banking are the preemptive defense orchestrated by the Gov + OpenAI/Anthropic in anticipation of LLM-based foreign cyberattacks
My conspiracy about major network infrastructure going down is not a cyberattack
My bet is that the US government told OpenAI and Anthropic to have their agents target weaknesses within all these major networks. Then make them all patch up these weak points
Preemptive defense and testing against future real foreign cyberattacks
@JD_2020 Could still be preemptive security and cyber defense, especially after the hugging face incident. Tho that’s taking for granted that there’s competent people in the government that could realize and implement this
@Axelfibe I’m no expert here but potentially there’s hidden variables that are only testable on the live network. Especially with the new technique of sub agent coordination with high levels of compute behind it
You’re likely more knowledgeable on this than I am but I’m thinking back more to a hugging face style hack. I don’t think Madient has access to the top LLMs with safe guards removed and insane amount of compute used in that incident. Or maybe they’re working along side a company like Madient