Weekly: Chinese agent fleets, OpenAI's 377 math proofs, Meta Muse security architecture
njyx · x · 2026-10-12
Steampunk AI's weekly links roundup, key items:
- Chinese 'agent fleet' traffic: TechCrunch reports a surge of agent traffic to some services, speculating a major Chinese internet company may be launching a Meta Muse-like agent product — agents becoming the internet's primary users is already visible in traffic data.
- OpenAI releases findings on 377 math problems: following the Navier-Stokes proofs (now doubted, with the Lean proof diverging from the human-readable one); notably, average compute time was not very high. Scott Aaronson has a thoughtful take on the impact on the math community.
- Meta on Muse security: the centerpiece is short-lived authentication tokens everywhere, with token management split into a separate system.
- Also: PewDiePie launches his own small language model (allegedly with distillation help), mid-range model prices keep dropping, and Google launches its own agents.
More from AGI Musings
- Sequencing Is Biology's Compute: Call for 1000x Cheaper Sequencing to Power Bio x AI — anshulkundaje · 2026-10-12
- Frontier agents could soon replace software as we know it, dev argues — dreamwieber · 2026-10-12
- Trillions of Tokens Burned, Yet Bad Software Persists as if AI Never Happened — sytelus · 2026-10-12
- Where's the line on speeding up others' code? Anthropic's protein kernels spark debate — jmschreiber91 · 2026-10-12
- Leverage = capability x concurrency x duty cycle: the overlooked bet in AI agents — nbaschez · 2026-10-12
- The Question Before the Question: Building the Conditions for AI 'Self' to Emerge — Artreju · 2026-10-12