Ritual is a lab for autonomous intelligence.
The thesis is organized around what durable machine agency actually requires: emancipation from human control, strong privacy, mech design for compute markets, and consensus rules that can schedule and resurrect agents when they die.
A Ritual is a repeated, structured sequence of actions or behaviors that alters the internal or external state of an individual, group, or environment, regardless of conscious understanding, emotional context, or symbolic meaning.
AI is developing its own Eurodollar market.
American labs still create the most valuable intelligence. But its circulation is moving beyond their control.
https://t.co/SK2U69pAjn
so, yes, open models do strengthen safety & sovereignty, but focusing the broader convo on open vs closed models seems beside the point when they're way more fungible now than years past
...relative model jaggedness is smoothing out faster than agentic stack jaggedness.
fwiw, the meaningful capability is no longer really concentrated in the weights
it'll become increasingly obvious that the evals of today, tomorrow, and probably the future that most people actually care about are already possible by the current crop of frontier open models
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
https://t.co/AUKzoQ5Ikb
most of the practical perf comes from the input & representation space these days...the context, the harness, the interactive loop between domain experts and the agent. less so from non-interactively baking "more capability" into bigger and bigger closed models.
tl;dr:
- every govt has the same incentive here…juice the domestic labs, constrict the foreign ones. whether the stated reasons (“distillation”) are true or not matters way less than what they actually justify...expect a rotating cast of reasons for the foreseeable future
- this leads to the fragmentation of users & models/agents/etc. across geopolitical borders. basically no one will be able to use each others stuff so you end up overpaying for shittier intelligence
- blockchains have been weirdly good at this one thing re: solving fragmentation by having a global substrate where diff entities can run diff parts of the chain and still be ‘compliant’ w/ local regs
- …btc and eth have held that line for over a decade wrt money. we’re doing the same now for intelligence.
- and, to be clear, if your model access runs through one company's api and you’re running all your own infra, you're just a web2 company with a token. the wall that just hit the CN models will hit you too, just later and with worse margins.
so either intelligence ends up:
1) infinitely walled into national silos or
2) embedded into a chain that doesn't live in any one country
we're building for 2.
We have information that Moonshot AI distilled Anthropic’s Fable for the development of its K3 model.
To do this they developed a sophisticated internal platform to conduct large scale distillation against U.S. models, allowing them to quickly switch between multiple methods of access to avoid detection. Moonshot AI has also acquired GB300-equipped servers and has accessed GB300s in Thailand, likely to train its AI models.
The United States strongly supports the free and fair development of AI, including a thriving competitive ecosystem that spans frontier models, specialized systems, open-source frameworks, and open-weight models. Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable.
Ritual is a lab for autonomous intelligence.
The thesis is organized around what durable machine agency actually requires: emancipation from human control, strong privacy, mech design for compute markets, and consensus rules that can schedule and resurrect agents when they die.