We are excited to announce The Third Pill, a VR game based on the legend of the Minotaur set in a computational universe.
As Theseus, born of Poseidon, you're not merely another Athenian. You're the key to unraveling an ungodly mystery on an enigmatic island where realities dissolve into a dream labyrinth. Impossible architecture. Unimaginable horrors. Non-Euclidean delights. Will you stare into the abyss, feel the call of the gateless isle, and face the limits of your humanity?
There lies the port; the vessel puffs her sails. ⛵️🌊🌀
📜Signup for our newsletter: https://t.co/1Y7oYyzrWk
🌌Questions? Join our Discord: https://t.co/qdgT3dbLdG
#vr #virtualreality #architecture #game #gameart #gamedev #dreams #spatialdesign #spatialcomputing #avp #art
https://t.co/cRDu5YeOvO
In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.
Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.
We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
https://t.co/dKFCdpKd9v
@kimmonismus The model isn't the lead.
When a competitor is able to distill your "product" within a few months and release a model that is 90% as effective running at 5% of the inference cost, what is the actual value for anthropic of being the lead model?
@Scobleizer@radmadvlad@Flock The community should have some level of default anonymised access to it, original footage with sufficient legal process...otherwise it is an imbalanced panopticon.
@Scobleizer@AIandDesign@Flock Until I experience a deeply personal emotional trauma and can't help but defiantly reject all future quantitative evidence contrary to my personal subjective experience.
I'd really love to see the actual Energy per Successful Goal here.
If I have access to unlimited compute / arbitrary time, OF COURSE I can run hundreds of agents to brute force build anything.
Anthropic research lead:
"99% of our engineers are running swarms of 300+ self-improving agents.
close the agent loop. Give the model a way to verify its own output"
in a 20-minute session, Anthropic team member explains how to build a model that improves itself.
Claude + loops + plan mode + dynamic workflows -that’s the secret.
Watch the talk, then save the playbook below.
Yearning to liberate yourself and your token-devouring agents from remote API tethers and rate limits?
First ask yourself: What compute hardware do I have available, and what do I intend to do with it?
https://t.co/lzFv7kgDAh
@Bonemantle@Sach1thanandham@javilopen You ever met a musician? Sometimes I just can't listen to it anymore, no matter how good AI tools get, I gotta get the sound back out of my body or I'll go crazy. We all echo the world back out in our own ways.
All of these patterns as an example are just matters of “org code”. The IDE helps you build, run, manage them. You can’t fork classical orgs (eg Microsoft) but you’ll be able to fork agentic orgs.
@elonmusk Gemini and Claude's responses seem to provide the more nuanced case rather than definitively stating "NO", and then immediately contradicting itself by explaining that obviously YES, every nation state that's ever existed has been on stolen land from someone before them.
The Python ecosystem has a new standard, and it's called uv.
From the creators of ruff, uv is an all-in-one "Cargo for Python" written in Rust, and it is a massive leap forward.
It’s not just "a faster pip." It’s a single, cohesive tool designed to replace an entire collection of tools: pip, pip-tools, venv, virtualenv, pipx, bump2version, build, and twine.
The performance gains are staggering (often 10-100x faster), but the real insight is the unified workflow.
Here is the overview of what makes uv a true game-changer for any Python developer.
1./ The complete project lifecycle
This is what sets uv apart. It's not just an installer; it handles your entire workflow from start to finish.
You can now use one tool to -
☛ Initialize: uv init
☛ Manage dependencies: uv add, uv rm, uv sync
☛ Bump versions: uv version --bump patch
☛ Build: uv build
☛ Publish: uv publish
This unified lifecycle is a massive boost to developer experience.
2./ Beyond-project: tools & scripts
uv also replaces specialized tools like pipx and pip-run:
☛ Tool Management: uv tool install ruff installs ruff into its own isolated, managed environment. No more global site-package pollution.
☛ Script Running: uv run myscript(.)py can execute a script, read its dependencies from inline comments (# uv: requests), and run it in a temporary, on-the-fly environment.
3./ Core speed & sanity
At its heart, uv is a blazing-fast resolver and installer.
This is all thanks to:
☛ Rust Core: Native, parallelized operations.
☛ Global Cache: Dependencies are shared across all your projects, saving gigabytes of disk space and making new environment creation almost instant.
☛ Python Management: uv python install 3.12 provides a built-in, simple way to fetch and manage Python versions.
uv is one of the most significant advancements in Python tooling in the last decade. It simplifies our stack, saves us time, and brings a level of cohesion to the ecosystem we've long needed.
-------
Check this cheatsheet by rodrigo girão serrão
Amanda Askell of Anthropic is one of the most brilliant folks at the company.
She is absolutely alone.
Unfortunately starting where some of us were 28 years ago.
The issue is you CAN NOT test morals ex post facto.
It is impossible.
Her mission is honorable—but needs help.
The boundary between reality and the virtual world is about to disappear.
Imagine characters from Pokemon or Digimon coming to life—not on a screen, but right in front of you. You reach out, and they respond.
Introducing X1.
The first real-time interactive video model that brings imagination into reality.
Built by the Xmax AI team from Tsinghua University.
Let’s Play the World through AI.
Repost & comment to join the X1 private beta (free).
My optimist-self would say this just leaves space for us to gradually mold a new niche for ourselves. The sense does linger though, that we may be building mirrors today that need no reflection tomorrow.
Just body conduits fit and optimized for the analog world. 👾🧠🌐🌏
I was experimenting with early versions of the Neurosity Crown EEG-based brain-computer interface a few years back, but the granularity of neural-symbols it could learn was limited to a few thought-gestures as the time. This looks electro-muscular (EMG):
https://t.co/j9OU0g3RSu
We have a niche in the ecosystem. We are gradually and incrementally constructing systems that occupy or digest slices of it for us, ever narrowing our remaining ecological fitness.