mostly wrong
Hell
Joined November 2021
- Tweets3.7K
- Following782
- Followers769
- Likes4.6K
Hope everyone is ready for when the internet gets deanonymized and they slap social credit scores on all of us.
People handing out almond joys on Halloween are liable for the flaming bag of shit that might get left on their porch.
Through exhaustive research (lighting money on fire) I’ve come to the conclusion the only things worth bidding for the foreseeable future are zec, cyph, and hype related.
What's the point of the corporate ladder if they keep adding rungs?
Everyone talks about how they’re building the next flywheel, but all I’ve seen so far is a bunch of wheels with wax wings.
What’s stopping AI from eventually sending most apps to zero when it can reverse engineer everything?
Maybe things do eventually get cheaper when there are no more moats.
Or at least we’ll get better customer service?
If the frontier model companies are so smart, why the fuck were they allowing internet access? Maybe AI isn't that smart, we're all just retarded.
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
If the internet keeps getting hacked (it will). It seems extremely positive for the most forward thinking security team (ZEC). Not sure anyone is ready for the repricing coming in the next few years.
Getting the urge to start calling out all of the retarded shit that seems to be popping up on my timeline.
Reports about the new generation being dumb are apparently accurate. Everybody knows it has to be a radioactive spider.
NEW: An 8-year-old boy let a black widow bite him, hoping to become Spider-Man.
Readers added context they thought people might want to know
The incident took place in Bolivia in 2023. The boy recovered after treatment with antivenom.
newsweek.com/boy-8-let-blac… mirror.co.uk/news/world-new…