1471
Expanding on what we missed with sycophancy
A deeper dive on our findings, what went wrong, and future changes we’re making.
New
The latest blog posts to take off on Hacker News. 2,111 posts.
1472
1473
Crawlers impact the operations of the Wikimedia projects
Since the beginning of 2024, the demand for the content created by the Wikimedia volunteer community – especially for the 144 million images, videos, and other files on Wikimedia Commons – has grow…
1474
Webflow makes GSAP 100% free – plus more updates
Discover exciting updates to GSAP — supported by Webflow — plus product enhancements and more.
1475
1476
Google Wallet launches new age and identity verification features (ZK proofs)
Learn more about new Google Wallet updates, including new ways to use your digital ID for age and identity verification.
1477
Oxide’s compensation model: how is it going?
Reflecting on how our compensation model is going
1478
New Study: Waymo is reducing serious crashes and making streets safer
The path to Vision Zero requires reducing severe crashes and improving the safety of those most at risk. Our latest research paper shows that the Waymo Driver is making significant strides in both areas. By reducing the most dangerous crashes and providing better protection for pedestrians, cyclists, and other…
1479
The future of solar doesn't track the sun
PV modules are cheap enough today that the simple fixed East-West arrays are cheaper and faster to install than the industry’s darling, the single-axis tracked array. Xavier Dedenbach, Terraform Po…
1480
Dopamine signals when a fear can be forgotten
Study shows how a dopamine circuit between two brain regions enables mice to extinguish fear after a peril has passed.
1481
Claude Integrations
Claude can now connect to your apps and tools through Integrations, while advanced Research searches across the web, Google Workspace, and connected services to quickly deliver comprehensive reports.
1482
We identified a North Korean hacker who tried to get a job
Cybersecurity isn’t just a priority for Kraken, it’s part of our origin story and embedded into our DNA.
1483
Vanguard 50-year anniversary CEO letter
Vanguard's thought leaders share timely insights about international financial markets, investor behavior, and the investment industry.
1484
AI code review: Should the author be the reviewer?
AI reviewing AI-generated code - genius or pointless? Deep analysis of AI code review paradox + real data on why it actually works better than humans.
1485
The Brief Origins of May Day
Most people living in the United States know little about the International Workers' Day of May Day. For many others there is an assumption that it is a holiday celebrated in state communist countries like Cuba or the former Soviet Union. Most Americans don't realize that May Day has its origins here in this country…
1486
Trust Me, I'm Local: Chrome Extensions, MCP, and the Sandbox Escape
Secure the modern software stack with Koi. Gain visibility, context, and real-time protection over extensions, MCPs, and autonomous AI agent endpoints.
1487
Running Qwen3 on your macbook, using MLX, to vibe code for free
Localforge blog post
1488
Strings Just Got Faster
In JDK 25, Strings used as keys in immutable Maps can be much faster.
1489
Phi-4 Reasoning Models
Microsoft continues to add to the conversation by unveiling its newest models, Phi-4-reasoning, Phi-4-reasoning-plus, and Phi-4-mini-reasoning. Learn more.
1490
108B Pixel Scan of Johannes Vermeer's Girl with a Pearl Earring
Girl with a Pearl Earring
1491
Espressif's ESP32-C5 Is Now in Mass Production
Today, we are glad to announce that ESP32-C5 is now in mass production.
1492
RustAssistant: Using LLMs to Fix Compilation Errors in Rust Code
The Rust programming language, with its safety guarantees, has established itself as a viable choice for low-level systems programming language over the traditional, unsafe alternatives like C/C++. These guarantees come from a strong ownership-based type system, as well as primitive support for features like closures…
1493
Mercury: Commercial-scale diffusion language model
We trained diffusion large language models that are up to 10x faster and cheaper than current LLMs, pushing the frontier of intelligence and speed for language models.
1494
Future of OSU Open Source Lab in Jeopardy
I am writing to inform you about a critical and time-sensitive situation facing the Open Source Lab (OSL). Over the past several years, we have been operating at a deficit due to a decline in corporate donations. While the Oregon State College of Engineering (CoE) has generously filled this gap, recent changes in…
1495
NotebookLM Audio Overviews are now available in over 50 languages
Learn more about NotebookLM’s Audio Overviews feature and its expansion to more than 50 languages.
1496
In kids, EEG monitoring of consciousness safely reduces anesthetic use
Using EEG readings of brain waves to monitor unconsciousness, an anesthesiologist can significantly reduce the amount of the anesthesia administered to safely induce and sustain the anesthetized state of children undergoing surgery.
1497
Sycophancy in GPT-4o
We have rolled back last week’s GPT‑4o update in ChatGPT so people are now using an earlier version with more balanced behavior. The update we removed was overly flattering or agreeable—often described as sycophantic.
1498
What It Takes to Defend a Cybersecurity Company from Today's Adversaries
This report highlights a rarely-discussed but crucially important attack surface: security vendors themselves.
1499
Waymo and Toyota outline partnership to advance autonomous driving deployment
Toyota Motor Corporation (“Toyota”) and Waymo reached a preliminary agreement to explore a collaboration focused on accelerating the development and deployment of autonomous driving technologies.
1500
How much energy does ChatGPT use?
This Gradient Updates issue explores how much energy ChatGPT uses per query, revealing it's 10x less than common estimates.