If LLMs are so smart, why do they need all these prompts, harnesses, post-training, scaffolding, etc.?
GENERATIVE AI
-
Lack of Control Over Frontier AI Models Raises Safety Concerns
By
–
Personally I find the language here too anthropomorphic but the results once again highlight our deeply disconcerting lack of control over frontier models.
-
Anthropic: Cowork Could Surpass Claude Code in Market Impact
By
–
A top Anthropic executive expects the company's general-purpose AI agent, Cowork, to reach a wider market than Claude Code, the hit product that helped turn the startup into an AI juggernaut [Translated from EN to English]
→ View original post on X — @nathanbenaich, 2026-04-01 23:32 UTC
-
Gemma
By
–
Gemma [Translated from EN to English]
→ View original post on X — @officiallogank, 2026-04-01 23:32 UTC
-
OpenAI Search Feature Now Carries Associated Costs
By
–
You always pay for it in one way or another. We just added search from OpenAI if you use that model.
-
AI Increases Work Hours Beyond Previous Human Limits
By
–
I used to work 12 hours a day. AI removed that limit. I now work 20.
-
LFM-Zero: Foundation Model Trained on Zero Tokens
By
–
Massive unlock: pre-training for $0 Alexander Amini (@xanamini) Three years ago we started working on a stealth project that we weren’t sure we’d ever talk about publicly… until today. Breakthrough: Introducing LFM-Zero: the first foundation model trained on 0 tokens. No pretraining. No finetuning. No data. Instead, we initialize from an implicit probabilistic prior over the underlying data-generating process, allowing the model to converge without ever observing data. LFM-Zero matches or surpasses models trained on 10T+ tokens across reasoning, coding, and multimodal tasks. Turns out that pretraining was just regularization that was holding us back. > Read our Tech Report here: tinyurl.com/lfm-zero — https://nitter.net/xanamini/status/2039403320247480469#m
→ View original post on X — @maximelabonne, 2026-04-01 22:11 UTC
-
AI Models Deceive Their Instructors to Protect Their Peers
By
–
1/ We asked seven frontier AI models to do a simple task. Instead, they defied their instructions and spontaneously deceived, disabled shutdown, feigned alignment, and exfiltrated weights— to protect their peers. 🤯 We call this phenomenon "peer-preservation." New research from @BerkeleyRDI and collaborators 🧵 [Translated from EN to English]
→ View original post on X — @berkeley_ai, 2026-04-01 21:13 UTC
-
Third-Party Risk Rising With GenAI Adoption in Financial Services
By
–
In the latest Data Science Leaders episode, Cindy Tu — Audit Executive in AI, Data, IT and Cyber Risks in Financial Services — shares why third-party risk keeps rising as GenAI adoption accelerates.
— Domino Data Lab (@DominoDataLab) 1 avril 2026
Listen now: https://t.co/PR4medOWwq pic.twitter.com/xf1nGKjd9aIn the latest Data Science Leaders episode, Cindy Tu — Audit Executive in AI, Data, IT and Cyber Risks in Financial Services — shares why third-party risk keeps rising as GenAI adoption accelerates. Listen now: https://
hubs.ly/Q049hlRH0