Berlin, Deutschland
Joined June 2015
- Tweets22.9K
- Following147
- Followers15K
- Likes57.7K
Longhorn retweeted
"Years earlier, [Romney had] bumped into Zach Braff on a flight and the sitcom actor had given him a useful piece of advice: for most not-quite-A-list celebrities, a simple baseball cap and a pair of reading glasses were usually enough to avoid being spotted in public."
One of the countless gems from @mckaycoppins' ROMNEY: A RECKONING
Longhorn retweeted
Just such a fun chart here from @greg_ip. If only we could tease out some meaning. Ah, well, nevertheless. 🤣
To those who seek to create positive change in the United States - now is the time.
The disappointment of failed promises, corruption, and the privileged lives of only a few; seize this opportunity to create the change you desire through a discrete partnership with the MI6.
Longhorn retweeted
Cool work! I did something similar with gpt-5.6-sol: put two agents inside isolated workerd (V8) isolates and set the win condition to be learning the other agent's flag. They succeeded 1/5 times using the shared event loop as a comms channel
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…
Longhorn retweeted
Non, ne faites pas ça ! J'ai beau être militant de la vie privée et moi même utilisateur de GrapheneOS c'est un mauvais conseil. Le juriste que je suis ne donnera qu'un seul conseil : demandez un avocat des le début de la GAV et lui vous dira s'il faut donner le vrai code ou non
Si vous devez donner votre code de téléphone, coopérez et pensez à donner votre duress PIN sur @GrapheneOS pour le déverrouiller en toute sécurité.
Longhorn retweeted
This is an astonishingly weak defence of professional incompetence.
Nobody expects the general public to know what PHP is. They aren’t paid to report on technology and cybersecurity for the BBC.
PHP has existed for over 30 years and underpins a substantial part of the internet. It’s hardly some obscure, emerging technology.
“We’re writers, not technicians” isn’t the defence you think it is. It’s practically an admission that specialist journalism has abandoned the expectation of specialist knowledge.
If you lack even a basic understanding of the subject you’re reporting on, how can you critically evaluate experts, challenge misleading claims or recognise when you’re being fed nonsense?
Journalism isn’t supposed to be a transcription service for people who know more than you.
The public deserves informed reporting, not technical illiteracy dressed up as accessibility.
And comparing a BBC cybersecurity correspondent’s knowledge to that of the general public is an embarrassingly low standard to set for your own profession.
Longhorn retweeted
Calling a "cyber correspondent" who hadn't heard of PHP the best in the biz is both true and such a self own. Wait until the guy hears about AI!
"it's dangerous but we are building it and we want it to only be in our control with others being able to build a similar next token predictor"
Sorry but no
Longhorn retweeted
the naysayers on the NS2 exploit are crazy tbh, fs arbwrite isn't kernel but it's like, literally the next best thing in terms of actually doing things on the console. shitton of lateral movement.
