Reposted by Charleshalvarflake.bsky.social @halvarflake.bsky.social · 17/09/2026I gave a talk at Bluehat Singapore today. The slides are here: thomasdullien.github.io/about/slides... 1219
Charles @charles.dardaman.com · 13/09/2026Even with all our errors and Ted Cruz cursing us we prevailed #hookem 141
Charles @charles.dardaman.com · 05/09/2026I wonder how correlated the focus of Astra on 3d envs and OpenAI’s larger robotic ambitions are 000
Charles @charles.dardaman.com · 05/09/2026When you know your request is about to be denied because it’s cyber adjacent 020
Charles @charles.dardaman.com · 04/09/2026The nice thing about AI hacking is that people are realizing just how insecure most software is 030
Reposted by CharlesJason Kikta @kikta.net · 13/04/2026DOLLY PARTON IS WEAK ON ADVANCED SEMICONDUCTOR EXPORT CONTROLS 10572116
Reposted by CharlesJoe Slowik @pylos.co · 08/02/2026“Ok we need a rousing version of the national anthem.” “Gotcha boss, real smooth jazz rendition, just like you asked for” “Wut?” 142
Charles @charles.dardaman.com · 03/02/2026Pretty strong comment in Nature today saying that AGI is here. "... the conclusion is straightforward: by reasonable standards, including Turing’s own, we have artificial systems that are generally intelligent." 110
Charles @charles.dardaman.com · 02/02/2026Management quantifies it by giving you a small raise and then denying all promotions as you’re too valuable to move up. 000
Reposted by CharlesJoe Slowik @pylos.co · 02/02/2026Propose we replace "spot the fed" at DEF CON (since there's so many of them anyway) with "six degrees to Epstein" this year 4222
Charles @charles.dardaman.com · 02/02/2026I can't see why this would ever be good to do. Dying is a natural part of life and figuring out how to overcome grief and move on is good. Prolonging that for years while trying to trick yourself into thinking they are alive through AI is a terrible way to handle it. 020
Charles @charles.dardaman.com · 01/02/2026So far there seems to be several different #infosec names in the Epstein files most of them after his first conviction. 010
Reposted by CharlesSimon Batzner @simonbatzner.bsky.social · 30/01/2026Our team at Google DeepMind is hiring for multiple roles in AI+materials. 🚀 We‘re pursuing an ambitious research program in materials and are looking for the world‘s best: - semiconductor experts - ml+materials researchers - lab automation engineers The vibes are great, join us! 12510
Charles @charles.dardaman.com · 29/01/2026Better to leave the controls you want in your product in place than remove them all and hope someone else follows the law which you can’t really judge without extreme oversight www.reuters.com/business/pen...reuters.comExclusive: Pentagon clashes with Anthropic over military AI use - sourcesThe Pentagon and artificial-intelligence developer Anthropic are at odds over potentially eliminating safeguards that might allow the government to use its technology to target weapons autonomously an... 041
Charles @charles.dardaman.com · 29/01/2026Copper is up 10% today and now I’m wondering if the empty house next to me really needs those wires in the walls 010
Charles @charles.dardaman.com · 29/01/2026Depending on the model size you might be surprised how much you can do while GPU poor, especially finetuning. 000
Charles @charles.dardaman.com · 29/01/2026I'm going to need to start screenshotting posts when quoting because people keep deleting them. 010
Charles @charles.dardaman.com · 29/01/2026Telsa's stock price is not based on current revenue it is based on his AI and robotics plays. Unless those fail, the revenue doesn't particularly matter 020
Reposted by CharlesQuantian @quantian.bsky.social · 29/01/2026Going up against Joe Weisenthal is a real coughing baby vs. hydrogen bomb moment from someone who’s about to find out they’re the baby 4781
Reposted by Charlesaria @aurelium.me · 27/01/2026openrouter.ai/arcee-ai/tri... we just released a preview version of Trinity Large (along with a base and pre-anneal checkpoint)! give it a try if you like Freeopenrouter.aiTrinity Large Preview (free) - API, Providers, StatsTrinity-Large-Preview is a frontier-scale open-weight language model from Arcee, built as a 400B-parameter sparse Mixture-of-Experts with 13B active parameters per token using 4-of-256 expert routing.... 1173
Charles @charles.dardaman.com · 28/01/2026Just in the last 8 weeks the models have gotten exponentially more useful. Seems pretty clear this is a step change in how people will work with code in the future. 010
Charles @charles.dardaman.com · 28/01/2026A significant amount of workers are using AI and then just not telling their bosses they are, so this stat doesn't surprise me at all. In fact the employees you want to keep probably want to cut out dull work with AI and the ones who don't are probably on your layoff list. 041
Charles @charles.dardaman.com · 27/01/2026Great to see an American company release a large open weight model #ai www.arcee.ai/blog/trinity...arcee.aiArcee AI | Trinity Large: An Open 400B Sparse MoE ModelA deep dive into Trinity Large, covering architecture, sparsity, training at scale, and why we shipped Preview, Base, and TrueBase checkpoints. 000
Reposted by CharlesThorne 🌸 @ens0.me · 27/01/2026It's good when technology makes it so that less toil is necessary to keep society functioning and advancing I believe this deeply and sincerely This is even an orthodox Marxist position 3784
Charles @charles.dardaman.com · 27/01/2026There's little reason to convince people of LLMs usefulness they can just be left behind. 150
Charles @charles.dardaman.com · 27/01/2026This is why their chat app is one of my least favorites even though codex is great 000
Charles @charles.dardaman.com · 27/01/2026Buying and scanning a book for internal use isn't theft. This is actually how it should be done in order to digitize it for LLM training whereas theft would be pirating it 010
Charles @charles.dardaman.com · 27/01/2026Love the focus on safety with this robot and the minimization of pinch points. I just hope they can get the cost down from the starting 50k so that someone would actually put it in their home. 061
Charles @charles.dardaman.com · 27/01/2026This looks immediately useful and a better value than paying overleaf since you probably already pay for OpenAI. Hopefully their terms of service aren’t terrible. #ai openai.com/index/introd...openai.comIntroducing PrismAccelerating science writing and collaboration with AI. 000
Charles @charles.dardaman.com · 27/01/2026If you know that all your PRs will be written by AI theres no point to accept them. Maybe theres some other future path such as someone giving tokens to pay for the AI to code up the feature they want instead. 000
Charles @charles.dardaman.com · 26/01/2026Being able to use AI to tell me what color things are is an underrated use of it. Now it’s a lot easier for me to figure out what smoke alarm battery is low 000
Charles @charles.dardaman.com · 26/01/2026Dario has written a companion essay to Machines of Loving Grace www.darioamodei.com/essay/the-ad...darioamodei.comDario Amodei — The Adolescence of TechnologyConfronting and Overcoming the Risks of Powerful AI 010
Charles @charles.dardaman.com · 26/01/2026This is obviously incorrect today. The value and ability for AI to be useful in white-collar work is apparent to everyone who uses it 000
Charles @charles.dardaman.com · 26/01/2026Microsoft is building their own chip for inference, I'd be curious how it compares to Google's TPU in practice. blogs.microsoft.com/blog/2026/01... 000
Charles @charles.dardaman.com · 24/01/2026This is why I more treat it like an enthusiastic junior who probably never learned what SQL injection was in college, but they can still pump code out. 000
Charles @charles.dardaman.com · 24/01/2026This would be great, but if AI becomes the entry point to the internet then people will start to SEO for AI in order to get traffic and it will be interesting to see what that looks like 030
Charles @charles.dardaman.com · 19/01/2026They will perfect this kind of application on remote workers under the guise of “catching North Korean workers” and information security 000
Reposted by CharlesCharles @charles.dardaman.com · 18/01/2026AI's increase in desire for software will be a boon for engineers who love to solve problems and work with product and be terrible for engineers who just love coding as thats going to quickly fall away into the background. 132
Reposted by CharlesEthan Mollick @emollick.bsky.social · 18/01/2026Erdos problems, a set of famous difficult math challenges, are a clear example of AI models breaching a threshold. The idea that an AI could solve one, let alone many, would have been insane a year ago (o1 was brand new). Now we have multiple Erdos problems solved by GPT-5.2 in the last couple weeks 513223
Charles @charles.dardaman.com · 18/01/2026AI's increase in desire for software will be a boon for engineers who love to solve problems and work with product and be terrible for engineers who just love coding as thats going to quickly fall away into the background. 132
Charles @charles.dardaman.com · 17/01/2026How are we to believe that no details are shared with advertisers while the example image shows ads that are highly relevant to the private conversation? #openai #ai #ads 010
Charles @charles.dardaman.com · 16/01/2026This is absolutely correct, the market of 2020 where anyone could get hired into Infosec is dead and you need to actually compete with a degree + experience and certs if you are trying to break in today. 000
Charles @charles.dardaman.com · 16/01/2026A MacBook SSHd into a Linux server remains the best way to work hands down 060