American in Norway. I’m trying my best!
Oslo
Joined March 2007
- Tweets35.9K
- Following820
- Followers17.8K
- Likes14K
Pinned Tweet
Everybody is worried about AI taking their job. I have looked at my inbox and I am not sure AI would want it.
Alfred Nobel's will put the Peace Prize in Norway. Nobody knows why. The fun theory: he did it to annoy Sweden, which he had complicated feelings about.
** cries in sovereign wealth fund **
🚨🇺🇸🇳🇴 Trump changes his tune on Norwegian migrants:
"I don’t want anybody from Norway. I don’t want anything to do with Norway.
I didn’t get the Nobel Prize, even though I ended wars."
Norway is one of the wealthiest and happiest countries in the world. I don't think they want to leave anyway. 🤷🏼♂️
Writer: Michael
Resurrected my Substack if you'd like to receive my musings in your inbox,
seanpercival.substack.com/
The most common question I get from Norwegians about America:
“Do they really drink booze in a brown paper bag like the movies?”
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…
At my job I keep getting invited to “alignment” meetings. Not sure if we just became a lot more corporate and this is a typical corp-slop word, or if I’m actually the problem/unaligned.
Oh no your screen is broken #̢̛ͥ̀̽̽ͨ̌̓͆̇̀̽ͦͨͫͯ̏̎̇̈̚͜҉̕͢͏͡҉̸̶̛͢͡҉̵̨̭̳̘̹̫͔͖̜̠͖̫̲̖̱̤̻̖͉͢͞͡ͅͅ͏̸̷̢͑̊̀̐ͣͭͭ̉͂ͨ̽͒̊̎͛ͪ̊̎ͩͧ̀͝͠͠͏̸̵̵̷̵