ice.computer
→
ML News
refresh
1.
▲
System Card: Claude Mythos Preview [pdf]
818 points
(www-cdn.anthropic.com)
7 days ago
2.
▲
Project Glasswing: Securing critical software for the AI era
1477 points
(www.anthropic.com)
7 days ago
3.
▲
GLM-5.1: Towards Long-Horizon Tasks
608 points
(z.ai)
7 days ago
4.
▲
Assessing Claude Mythos Preview's cybersecurity capabilities
307 points
(red.anthropic.com)
7 days ago
5.
▲
Claude Code login fails with OAuth timeout on Windows
222 points
(github.com)
7 days ago
6.
▲
Show HN: Gemma 4 Multimodal Fine-Tuner for Apple Silicon
222 points
(github.com)
7 days ago
7.
▲
AI may be making us think and write more alike
227 points
(dornsife.usc.edu)
7 days ago
8.
▲
Taste in the age of AI and LLMs
262 points
(rajnandan.com)
7 days ago
9.
▲
MegaTrain: Full Precision Training of 100B+ Parameter LLMs on a Single GPU
249 points
(arxiv.org)
6 days ago
10.
▲
Show HN: Output.ai - OSS framework we extracted from 500+ production AI agents
40 points
(output.ai)
7 days ago
11.
▲
Claude Managed Agents
137 points
(claude.com)
6 days ago
12.
▲
OpenAI says its new model GPT-2 is too dangerous to release (2019)
377 points
(slate.com)
6 days ago
13.
▲
AI Assistance Reduces Persistence and Hurts Independent Performance
17 points
(arxiv.org)
6 days ago
14.
▲
Show HN: We fingerprinted 178 AI models' writing styles and similarity clusters
73 points
(rival.tips)
6 days ago
15.
▲
Show HN: TUI-use: Let AI agents control interactive terminal programs
36 points
(github.com)
6 days ago
16.
▲
Muse Spark: Scaling towards personal superintelligence
251 points
(ai.meta.com)
6 days ago
17.
▲
Muse Spark – Meta Superintelligence Labs
191 points
(meta.ai)
6 days ago
18.
▲
ML promises to be profoundly weird
355 points
(aphyr.com)
6 days ago
19.
▲
GLM-5.1 matches Opus 4.6 in agentic performance, at ~1/3 actual cost
20 points
(app.uniclaw.ai)
6 days ago
20.
▲
MemPalace, the highest-scoring AI memory system ever benchmarked
58 points
(github.com)
7 days ago
21.
▲
The highest-scoring AI memory system ever benchmarked
21 points
(github.com)
7 days ago
22.
▲
Google open-sources experimental agent orchestration testbed Scion
225 points
(www.infoq.com)
7 days ago
23.
▲
Show HN: Skrun – Deploy any agent skill as an API
41 points
(github.com)
6 days ago
24.
▲
Anthropic's Project Glasswing sounds necessary to me
55 points
(simonwillison.net)
6 days ago
25.
▲
LLM scraper bots are overloading acme.com's HTTPS server
64 points
(acme.com)
6 days ago
26.
▲
Show HN: Finalrun – Spec-driven testing using English and vision for mobile apps
27 points
(github.com)
7 days ago
27.
▲
Show HN: Marimo pair – Reactive Python notebooks as environments for agents
27 points
(github.com)
7 days ago
28.
▲
No \"New Deal\" for OpenAI
18 points
(minutes.substack.com)
7 days ago
29.
▲
Claude Is Not Your Architect. Stop Letting It Pretend
18 points
(www.hollandtech.net)
7 days ago
30.
▲
Scientists invented a fake disease. AI told people it was real
13 points
(www.nature.com)
6 days ago
31.
▲
The pinnacle of enshittification, or Large Language Models
13 points
(blogs.gentoo.org)
6 days ago
32.
▲
AI Did It in 12 Minutes. It Took Me 10 Hours to Fix It
13 points
(idiallo.com)
5 days ago
33.
▲
New York Times Got Played by a Telehealth Scam and Called It the Future of AI
35 points
(www.techdirt.com)
6 days ago
34.
▲
The AI Great Leap Forward
74 points
(leehanchung.github.io)
6 days ago