0. Meta Ignores Brazilian Court Order to End Bolsonaro Favoritism (panamericandispatch.org)
474 points · 329 comments · by vrganj
Sleeping Giants Brasil and CTRL+Z submitted evidence alleging Meta’s platforms favor Flávio Bolsonaro and link Lula to derogatory content, prompting Brazil’s electoral court to demand explanations, enforce a potential R$500,000 daily fine, and consider doubling it for continued noncompliance. [src]
1. A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis (lynchlab.net)
506 points · 228 comments · by sungam
A dermatologist’s test of 200 biopsy-proven lesions found a fine-tuned Inception v3 CNN correctly diagnosed 62%, outperforming two frontier LLMs at 42%, though the small study’s results have substantial caveats. [src]
2. I refuse to accept a censored Internet. Join Me at FOSDEM 27 (talkweb.eu)
436 points · 210 comments · by Tomte
FOSDEM 2027’s Digital Dissident Devroom invites developers, researchers and activists to propose talks on integrated tools for combating censorship, protecting at-risk users and preserving digital freedom, with submissions open until November 30, 2026. [src]
3. Dolly Parton's Imagination Library (imaginationlibrary.com)
485 points · 87 comments · by jumploops
Dolly Parton’s Imagination Library promotes children’s dreams by providing books that help plant the seeds of learning and future aspirations in communities worldwide. [src]
4. Oregon Police Are Charging a Local News Outlet $2.5M for Records About One Cop (theintercept.com)
444 points · 121 comments · by tdeck
Eugene, Oregon, police estimated it would cost local outlet Double Sided Media nearly $2.5 million to process records about an officer accused of misconduct, prompting the outlet to appeal the fee as an intimidation tactic. [src]
5. AI use leads UC Berkeley math professor to see 'no point' in grading homework (dailycal.org)
212 points · 308 comments · by romanhn
UC Berkeley Fields Medalist Richard Borcherds has stopped grading homework for correctness, saying widespread AI use makes the practice futile, while the university leaves classroom AI policies to individual faculty. [src]
6. OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? (minimaxir.com)
238 points · 261 comments · by minimaxir
OpenAI and Anthropic slashed LLM prices by up to 80% as more efficient models, cheaper open-weight competitors, abundant inference providers and escalating rivalry pressured them to compete, while potential IPOs and expanding agentic usage added incentives to demonstrate scalable demand. [src]
7. Cerebras Deploys Trillion-Parameter AI Models (cerebras.ai)
355 points · 132 comments · by gmays
Cerebras explains how it distributes a 3-trillion-parameter Kimi K3 model across 59 WSE-3T wafers, using pipeline parallelism, on-chip memory, and KV-cache allocation to support high-throughput inference and concurrent requests. [src]
8. Serbia's Vucic concedes election defeat in historic win for student movement (reuters.com)
262 points · 145 comments · by consumer451
Serbia’s student-led protest movement won a historic parliamentary election, prompting President Aleksandar Vucic to concede after 12 years in power, though uncertainty remains over how its largely unknown candidates will form a government and navigate domestic and foreign policy. [src]
9. JetBrains Thinkrail (github.com)
200 points · 147 comments · by hmokiguess
JetBrains’ open-source ThinkRail is an agentic IDE built around the pi coding agent, offering isolated Git workspaces, living project specifications, concurrent chats, extensible tools, MCP support, and desktop or browser-based access. [src]
Brought to you by ALCAZAR. Protect what matters.