2026-09-17
Agents let AI safety share experiments hourly, not just papers monthly
Summary: Today, AI safety research is shared primarily at the scale of papers, creating collective feedback loops that take weeks or months. I propose an agent-based research approach that also...
Measuring alignment drift via trajectory prefixes
This work was done as part of MATS 10.0 under Maksym Andriushchenko. We present intermediate results here while we run further experiments.SummaryWe study alignment drift by asking LLM agents to...
TinyKPI
Your business analytics directly in your notch Discussion | Link
Constraining the capacity of physical side channels for AI verification and security
Based on recent events, including Dario Amodei's essay on pacing the frontier and the subsequent response, a coordinated slowdown in AI capability developments is now in the Overton Window (however...
Can parts of the HuggingFace incident be simulated?
TL;DR The following is an exploratory experiment about unintended cooperation of agents via unauthorized channels. Agents ran in isolated environments given a task that can't be completed without cooperation. The...
The Latter Days of Magic (1)
Note: Trying out a serial fiction thing.we know that restraint is the existence of multiple attractors and we observe that physics is the origin of all meaning which imply that...
Bolt Forge
Bolt's new agent with open-source models and 50X usage Discussion | Link
Higgsfield API
One async API for 50+ generative media models Discussion | Link
datasette 1.0a40
Release: datasette 1.0a40 Same security fix as 0.65.5, plus some neat new features and bug fixes: Plugins can now launch and manage background tasks using the new datasette.add_background_task() method. Thanks,...
datasette 0.65.5
Release: datasette 0.65.5 Security fix for an issue where a trailing newline in a requested table name could bypass table permissions and expose private rows, reported by dpfkdlemtp in GHSA-h547-rmjf-5m2m....
The Loss of Singularity
Written for Proof and Prompts. For context: Proof and Prompts is a communal blog for mathematicians’ thoughts on how AI is changing mathematics.This essay is not a high-level reflection on...
30 Years Ago, a Japanese Sci-Fi Fantasy Spinoff Quietly Transformed RPGs Forever
Modern Atlus RPG releases might not be recognized worldwide without one amazing game that launched 30 years ago.
If a Mac mini is agentic overkill, try this glowing pyramid that runs personal bots
'Intern 2' offers an isolated environment and cloudy inference to automate your world
Australia’s Collins-Class Submarine Have No Choice But To Keep Going – No Matter What
Australia set four requirements in 1978: suited to Australasian waters, advanced combat systems, supported by Australian industry, usable in peace and war. It wanted ten boats and could afford six....
The Most Important City in AI Safety May Be Singapore
万里长桥 - AIS Collaboration and Field-building to Bridge the East/West DivideSingapore is wonderful! I just returned from vacation there visiting a friend. What a beautiful city, delicious food, and friendly...
What the Huggingface Incident tells us about multi-agent interactions?
Edited with feedback from Advait Yadav.In a Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident, METR and Redwood Research offer us...
Launching RESI, the Institute for Responsible Superintelligence
In the famous marshmallow challenge, teams race to build the tallest tower out of marshmallows, dry spaghetti, tape, and string. The celebrated result is that kindergarteners, using trial and error,...
How to Promise and How to Keep Promises from Breaking
We are creatures fragmentary but indivisible, each of us a whole that is always simpler than its parts, but complicated by the struggle for a living unison between them. Nothing...
Ride the Rails on a OneWheel
The US has tons of abandoned rail, especially in urban environments, so it seems a shame to let it go to waste. [OneWheel] decided to modify one of their boards...
Welcome to Postreads
Discover and follow the best content from across the web, all in one place. Create an account to start building your personalized feed today and never miss out on great reads.
Tiny Generals
Asynchronous hex-based strategy. Command armies, capture bases, and outthink opponents at your own pace.
Play NowSupport Postreads
Enjoying the service? Help me keep it running and improve it further by buying me a coffee!
Buy me a coffeeContent Timeline
115,691 articles since 2008
Trending Now
Top 5 clicked items this week out of 1,600 added this week
Freshly added
New feeds to discover
