audaciter. building things that build things at https://xitter.cf/t.co/VFlGoLwdkc

USA
Joined February 2009
OpenAI CEO Sam Altman: "I don't even know a dramatic enough word to use… models can find bugs that are outside of the normal threat model of what we think a hacker can do… hardware-level, insane stuff… I no longer think we're gonna be able to just patch all the bugs."
45
Feels like OAI and Ant both are making moves to expand the capabilities of developers building on/around the models rather than eating dinner alone. Very good.
38
you telling me we trusted this guy with compression and expected anything less than http2shell? patch ya fetch folks
699
Cloudflare about to have a hell of a year
JUST IN: Amazon CEO Andy Jassy reveals AWS cloud computing capacity is sold out through 2027 amid surging AI demand.
40
vitromancer retweeted
we are going to have an extremely secure internet in 3-5 years and it will probably suck until then
9
15
3
256
11,198
vitromancer retweeted
11
5,347
123
28,678
276,705
My entry for the Cloudflare's "Build the next Github" competition: Forge. Powered by a microkernel, the forge UI and CLI tools work with MANY git providers, including GH and Gitlab, while offering self-hosted solutions on Cloudflare Artifacts, DOKS, and others. Really excited to open source what I've been working on for the past 3 years, and this is the first of many announcements I'll be making this week.
3
5
263
Thrusters are cool but the design is excellent. Def want a roadster.
41
We open sourced Void v1.1.11 and we made a music video for it. void.cloud
73
127
61
1,075
115,978
Me like Argon
1
1
45
vitromancer retweeted
AI sovereignty is not a zero-sum game, but many governments now believe it is. Cloudflare's answer: more local open-source models, model-agnostic security tools, and a commitment to giving nations genuine choice. cfl.re/4zbPv0H #BirthdayWeek
6
10
2
74
10,424
vitromancer retweeted
Replying to @BaT_MaN_23
Done.
91
900
59
7,314
288,513
vitromancer retweeted
10/2/2026 🧇 👽
392
3,599
839
23,919
1,243,781
Gait development mechanics win every time.
Can an LM, starting from random init (!!), learn to generate all of its pretraining data? Introducing Self-Play Pretraining with Zero Data. Two models start from random initialization: a generator proposes programs for a universal Turing machine and a learner trains on their outputs. We never train on any real data, but see predictable scaling on natural datasets: zero-shot val loss on images, text, audio, and melodies decreases predictably with self-play compute. And the learner develops in-context learning capabilities. A fun proof-of-concept, co-led with @AdityaCowsik and @KfirDolev and co-authors @gbruno_dl, @ANourya @noahdgoodman, and @YoavLevine.
249
Something deorbiting west of AZ? What is this? Not planes, slow moving, no trail but distinct tails. There were 3 but one became a fixed point of light and faded...
75