Alignment faking in large language models
A paper from Anthropic's Alignment Science team on Alignment Faking in AI large language models
Every blog post that took off on Hacker News, ranked by the votes of its best submission. 31,677 posts.
A paper from Anthropic's Alignment Science team on Alignment Faking in AI large language models
With the Deutschlandticket, you can travel across Germany by bus and train. Here you can find out everything need to know about it.
The software update includes a more unified user experience across connected devices, enhancements to AI Vision Inside™, expanded Knox Security and more
Access to this page requires authorization. You can try signing in or changing directories .
As market surveillance authority for radio equipment and responsible for controlling public exposure to electromagnetic fields, the ANFR runs inspections on mobile phones placed on the French market. 141 mobile phones, including Apple’s iPhone 12, have recently been tested to check compliance with limit SAR values…
Now the Fact Check label in Google News is available everywhere, and has been expanded into Search globally in all languages.
One year from now, with the release of Chrome 154 in October 2026, we will change the default settings of Chrome to enable “Always Use Secure Connections”. This means Ch…
There are tons of blog posts on the internet about how frameworks differ and which one to pick for your next web project. Usually they cover a few aspects of the framework like syntax, development setup, and community size. This isn't one of those posts.
News and insights on the Android platform, developer tools, and events.
This report is the first independent expert application of the 1948 Genocide Convention to the ongoing treatment of the Uyghurs in China.
When Best Practices Are Best
Opera Developer with native ad blocker and power saver enabled is able to run 22% longer than Microsoft Edge.
Information regarding Nintendo's rollout of the Nintendo Network shutdown
Tor, like much of the world, has been caught up in the COVID-19 crisis. Like many other nonprofits and small businesses, the crisis has hit us hard, and we have had to make some difficult decisions. We had to let go of 13 great people who helped make Tor available to millions of people around the world. We will move…
It appears that the clang-11.0.0 compiler mis-compiles sqlite3.c version 3.32.1. I have checked in a change to SQLite that appears to work around the problem. But there might be other bugs. Therefore, use clang-11.0.0 with caution and test your applications carefully!
Our TimescaleDB vs Amazon Timestream results surprised us, but after intensive testing, we found Timestream slow, expensive, and missing key capabilities.
Gain full control over IP address allocation in your Tailscale network, enabling efficient management and compatibility with other systems.
I've had a few conversations recently where I say things like, "the Japanese Ruby community uses Ruby for different things than in America"... and I get blank stares. Specifically, I mention that America is very centered on Rails and web apps with Ruby. No surprise, right? "But then,
Searching code at scale without employing a search team is hard
The state of the network after one year of operation.
The subtle ways in which design tools shape how we think and what we make
From the Global South to the Arctic, can machine learning-enabled weather prediction better protect communities and economies?
The New York Times' editorial board has made a disappointing endorsement of the Trans-Pacific Partnership (TPP), even as the actual text of the agreement remains secret. That raises two distressing possibilities: either in an act of extraordinary subservience, the Times has endorsed an agreement...
How we took eggs out of the S3 basket.
If you work in a newsroom, there’s a good chance you work with colleagues on Google Docs, Slides, Sheets, and more. Google Workspace software is simple and powerful. But anyone working in a newsroom has probably asked themselves: What can Google see? What about our most sensitive conversations and documents? What…
Private Cloud Compute (PCC) fulfills computationally intensive requests for Apple Intelligence while providing groundbreaking privacy and security protections — by bringing our industry-leading device security model into the cloud. To build public trust in our system, we’re making it possible for researchers to…