Good morning. It’s Monday, September 23rd.

Did you know: On this day in 2008, the first commercially available Android phone, the HTC Dream (also known as the T-Mobile G1) was announced.

In today’s email:

• Pudu Robotics

• OpenAI’s Secret Project with Johnny Ive

• Meta Bullying Open-Source Community?

• 4 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

Meet the Semi-Humanoid Redefining Service Robotics

Pudu Robotics has launched the PUDU D7, a 'semi-humanoid' robot designed to handle serving, sorting, and stocking tasks in the food industry. Standing 165 cm tall and weighing 45 kg, the robot has 30 degrees of freedom, which can be expanded to 50 with an additional hand attachment. It can lift up to 10 kg and operates for 8 hours on a single charge, matching the length of a typical workday. Tailored for service and industrial environments, this latest release reflects Pudu's continued expansion of its robot lineup. Read more.

Related Story: AnySkin Brings Touch to Robotics

iPhone designer Jony Ive and OpenAI might try to build the hardware for a real-life "Her"

Former Apple design chief Jony Ive is collaborating with OpenAI to develop an innovative AI device that prioritizes voice interaction over a traditional screen interface. Although details are limited, Ive has confirmed the acquisition of a 32,000-square-foot office space in San Francisco, where his design firm, LoveFrom, will lead the project.

This initiative has attracted a small team, including former Apple designers, focused on harnessing generative AI for tasks such as summarizing news and booking travel. OpenAI’s goal is to transition from conventional chat systems to more agentic AI solutions, inspired by the AI assistant Samantha from the film Her. Initial funding comes from Ive and Laurene Powell Jobs' Emerson Collective, with potential plans to raise up to $1 billion from investors like SoftBank, strategically positioning OpenAI in a competitive AI market. Read more.

Meta is accused of “bullying” the open-source community

Meta faces backlash from the open-source community for its approach to AI development, accused of "open washing" its AI models, particularly the Llama series. Critics argue that Meta's definitions and practices undermine the essence of open-source principles.

The Open Source Initiative (OSI) recently introduced draft standards mandating transparency in training data, source code, and model weights, which Meta's offerings do not fulfill. While Meta releases model weights, it withholds crucial training data and imposes licensing restrictions, failing to meet OSI criteria. This has led to concerns about potential exploitation of regulatory loopholes, particularly in light of the EU AI Act. Meta defends its stance, asserting the necessity for a flexible interpretation of openness. Read more.

Plan Like a Graph: New prompting method helps language models plan 

Study questions benefits of LLMs large context windows

China’s leading AI start-ups eye fresh opportunities after OpenAI previews latest LLM 

Microsoft’s Copilot key will be able to launch apps on Windows 11 soon

Three Mile Island nuclear plant will reopen to power Microsoft data centers

Samsung and TSMC have reportedly discussed building AI chip “megafactories” in the UAE

Google CEO Sundar Pichai announces $120M fund for global AI education

📄 Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolution

📄 3DTopia-XL: Scaling High-quality 3D Asset Generation via Primitive Diffusion ​​

📄 Imagine yourself: Tuning-Free Personalized Image Generation

📄 MuCodec: Ultra Low-Bitrate Music Codec

📄 Training Language Models to Self-Correct via Reinforcement Learning

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, apply here.