🤖 China goes viral with AI humanoid showcase

During the China’s hugely popular Spring Festival Gala, humanoid robot weren’t in the background; they were the stars. Now the clip is going viral worldwide, sparking excitement and big questions about what comes next. Let’s dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: China showcased humanoid robots on its biggest TV show.

China’s CCTV Spring Festival Gala, watched by most of the country, featured humanoid robots front and center. Unitree robots performed complex martial arts routines, including drunken boxing moves and self recovery after falls. Noetix and MagicLab robots joined actors and dancers, showing coordination with humans. Behind the spectacle sits China’s AI plus manufacturing push, with IPO plans and strong state backing accelerating deployment.

The showcase video is going viral across X and YouTube, gaining millions of views. Take a few minutes to check them out. You’ll be surprised.

When a humanoid can stay balanced, recover from a fall, and move safely right next to people on live TV, it’s showing the kind of uptime and control you need for picking, packing, basic assembly, inspection, and tool handling in real sites. China is stacking attention, policy, and capital in one funnel, with IPO plans and forecasts like 28,000 humanoids sold in China in 2026 pointing to real pilots moving beyond labs and demos. For manufacturers, that means cheaper trials of “general” robot labor in messy workflows where fixed automation still struggles.

Stop losing nuance when you type prompts. Wispr Flow captures your spoken reasoning, removes filler, and formats it into a clear prompt that keeps examples, constraints, and tone intact. Drop that prompt into your AI tool and get fewer follow-up prompts and cleaner results. Works across your apps on Mac, Windows, and iPhone. Try Wispr Flow for AI to upgrade your inputs and save time.

Try Wispr Flow

Evolving AI: Pentagon pressures Anthropic over AI limits.

The Pentagon is reportedly considering cutting ties with Anthropic after months of talks over how its Claude model can be used on classified networks. U.S. officials want top AI firms, including OpenAI, Google and xAI, to allow military access for all lawful purposes, spanning weapons development and battlefield operations. Anthropic has pushed back on removing limits tied to fully autonomous weapons and large-scale domestic surveillance.

The Pentagon wants top AI models running on classified networks with far fewer refusals, and it’s pressuring vendors to sign up for “all lawful purposes.” If that becomes the standard contract language, limits on fully autonomous weapons and mass domestic surveillance risk turning into negotiable vendor settings. Vendors that say yes get budget, distribution, and deep integration into intel triage, mission planning, and battlefield workflows. The reported Claude use via Palantir during the Venezuela operation shows how quickly “policy debates” can turn into operational deployment.

Evolving AI: Claude Sonnet 4.6 levels up coding and computer use.

Claude Sonnet 4.6 replaces 4.5 as the default model for Free and Pro users, keeping pricing at $3/$15 per million tokens. Early developers report stronger instruction following and more consistent coding output, often preferring it over the earlier Opus 4.5. The model handles longer context, complex spreadsheets, multi-step web forms, and cross-tab workflows with fewer errors. On OSWorld, which tests real software like Chrome and VS Code without special APIs, Sonnet shows steady gains in real computer interaction.

Sonnet 4.6 becoming the default for Claude Free and Pro means more people get near-top tier output without a plan upgrade, at the same $3/$15 per million token API pricing. The real kicker is computer use: teams can hand off busywork in legacy web apps, spreadsheets, and multi-step forms without building custom connectors. Pair that with the 1M-token context window in API beta for huge docs and codebases, and you can run longer end-to-end work in one thread. Prompt-injection tricks still exist on the open web, so guardrails and review stay part of the workflow, even with stronger resistance.

Evolving AI: Google’s 7 predictions on AI, LLM, and Observability.

Read the 7 key takeaways from Google’s Director of AI and Datadog’s VP of Engineering as they break down their predictions of the future:

Free Download →

QUICK HITS

🤖 Wordpress introduces AI assistant.

📺 Amazon’s Fire TV is now rolling out in the US.

🌎 Claims that AI can help fix climate dismissed as greenwashing.

🎓 University of Michigan student accused of using AI for college papers sues for disability discrimination.

 *partner link