ls /news · home news
grep -r --recent /news | head -20

News

50803 articles page 68 of 2541 0 sources 30 min sync cycle

// latest articles 50803 indexed

10:00
2026-07-14
opensourceceo.com
ai-policy · 14m read ↓ neg

Building A Second Brain With Claude Cowork

Apple sued OpenAI in federal court, alleging a coordinated scheme to steal its hardware secrets through a former engineer who left with confidential documents and coaching from OpenAI's hardware boss, who is also an ex-A…

09:56
2026-07-14
forum.effectivealtruism.org
artificial-intelligence · 5m read ↓ neg

Mistakes in AI 2027, Part 2

A critic's ongoing review of the AI 2027 report reveals that the authors misrepresent scientific sources, claiming evidence of AI reasoning and AI training AI where none exists. The referenced paper on scaled supervision…

09:54
2026-07-14
machinebrief.com
artificial-intelligence · 2m read ↓ neg

Can AI Rediscover Scientific Breakthroughs? Not Quite Yet.

A new benchmark called FIRE-Bench reveals that autonomous agents powered by large language models struggle to replicate recent high-impact scientific findings, achieving less than 50 F1 score. The results highlight AI's …

09:54
2026-07-14
machinebrief.com
artificial-intelligence · 2m read · neu

AI Routing: Precision Over Guesswork

A new benchmark derived from WildChat introduces a 12-agent catalog and 3,000 prompts to optimize accuracy-cost trade-offs in AI routing from natural-language prompts. Supervised routers, particularly fine-tuned encoders…

09:54
2026-07-14
arxiv.org
large-language-models · 2m read · neu

The Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge

A new study from computer scientists reveals that biases in LLM-as-judge scoring can be traced to specific geometric structures in the model's hidden state activations. The researchers identified low-dimensional subspace…

09:54
2026-07-14
machinebrief.com
artificial-intelligence · 2m read · neu

Can Your Text-to-Image Models Handle a Real-World Tweak?

Researchers developed the SPQR benchmark to test the reliability of safety measures in text-to-image diffusion models after fine-tuning. The benchmark evaluates safety, prompt adherence, quality, and robustness across mu…

← prev page 68 / 2541 next →
LIVE [news] indexed:50803 page:68/2541 en · ua 2026-05-20 ·