Tien Kung Ultra: Redefining endurance, one robotic stride at a time

China’s Tien Kung Ultra humanoid robot completed a 21km half-marathon in about 2 hours and 40 minutes, showcasing advancements in AI-driven robotics while Meta is enhancing Teen Accounts with AI-powered age verification and automatic protective settings to ensure safer online experiences…

All this and more in today’s newsletter

News Bytes

• China’s Tien Kung Ultra humanoid robot, developed by the Beijing Humanoid Robot Innovation Centre, won the country’s first half-marathon for humanoid robots. The 180cm, 55kg robot completed the 21km race in about 2 hours and 40 minutes, with engineers swapping its battery three times to optimize performance. The center, backed by major Chinese robotics firms, aims to create a foundational platform for humanoid robotics, likening its approach to Google’s Android for AI-enabled machines.

• Meta is expanding its Teen Accounts initiative to ensure more teens benefit from built-in protections across Instagram, Facebook, and Messenger. These accounts automatically limit who can contact teens, the content they see, and the time they spend online. Parents are notified on Instagram with guidance on verifying their teens' ages, and Meta is testing AI technology to identify suspected teen accounts and place them in protective settings, even if they list an adult birthday.

• Cursor’s AI support bot, “Sam,” invented a non-existent company policy, confidently claiming that users could only log in from one device at a time. This misinformation led to user outrage, subscription cancellations, and widespread complaints on forums like Reddit and Hacker News. Cursor later clarified the error, apologized, and refunded affected users, but the episode underscores the potential business damage caused by AI “hallucinations”—when chatbots fabricate plausible-sounding but false information.

• Anthropic’s research paper “Values in the Wild” investigates how their AI model, Claude, expresses values during real-world user interactions. By analyzing over 300,000 anonymized, subjective conversations, the team developed a taxonomy of values that Claude demonstrates, such as helpfulness, honesty, and harmlessness, and found that the model generally aligns with these prosocial ideals.

• MIT researchers have developed a new technique to improve the accuracy of AI-generated code by guiding large language models (LLMs) to produce outputs that strictly adhere to programming language rules. Their method enhances computational efficiency by discarding invalid outputs early, allowing smaller models to outperform larger ones in structured code generation.

AI Automation Bytes

• Make : From tasks and workflows to apps and systems, build and automate anything in one powerful visual platform.

• Zapier : Use Zapier to tame the chaos with automation and accomplish more with less work.

• HubSpot : AI-powered customer platform with all the software, integrations, and resources you need to connect your marketing, sales, and customer service.

• Scale AI : Scale Data Engine leverages your enterprise data, and with Scale GenAI Platform, safely unlocks the value of AI.

• Bardeen AI : Generate a custom automation for you when it detects manual tasks.

Start learning AI in 2025

Keeping up with AI is hard – we get it!

That’s why over 1M professionals read Superhuman AI to stay ahead.

• Get daily AI news, tools, and tutorials

• Learn new AI skills you can use at work in 3 mins a day

• Become 10X more productive

Sign up and start mastering AI

Feedback Bytes

Thanks for reading and If you’d like to sign up for this newsletter or share it with a friend or colleague, you can find us right here. 📬