ENFR
8news

Tech • IA • Crypto

TodayVideosVideo recapsArticlesTop articlesArchives

OpenAI AI device with Jony Ive, Google Cosmo leak, China robotics surge

AISunday, May 3, 2026· 8 videos

Briefing

0:00 / 0:00

OpenAI taps Jony Ive for device

OpenAI is reportedly developing a new AI-native device with designer Jony Ive, targeting a 2028 launch. The product aims to replace smartphones with a voice-first interface and no traditional screen. Qualcomm is expected to co-design the chip powering the system. The device would offload app interactions to an AI agent, redefining how users access digital services.

Google Cosmo app briefly leaks

Google accidentally published and quickly removed an experimental Android app called Cosmo. The app combined Gemini Nano on-device processing with cloud AI in a hybrid system. It also used accessibility APIs to read on-screen context, pointing to deep OS-level integration. The leak suggests Google is accelerating work on next-gen AI assistants.

Gemini Omni hints at unified model

Google is reportedly developing Gemini Omni, a unified media model that merges video and image generation. The system could replace separate tools like Veo and other image models. Early interface appearances suggest a possible reveal around Google I/O 2026. This signals a push toward all-in-one multimodal generation platforms.

China dominates robotics and patents

China has overtaken global competitors in both industrial robotics deployment and patent filings. It now operates more robots than the G7 combined, reflecting a long-term industrial strategy. The country has shifted from low-cost manufacturing to advanced technology leadership. This transformation is reshaping global supply chains and competitive dynamics.

Europe faces deep tech dependency

Europe is increasingly reliant on Chinese supply chains, particularly in hardware ecosystems like Shenzhen. Decades of offshoring weakened domestic industrial capacity. Even with reshoring efforts, critical components remain tied to Asian networks. This creates structural vulnerability in sectors like robotics and advanced manufacturing.

Demis Hassabis pushes scientific AI vision

Demis Hassabis of DeepMind is advocating for AI focused on irreversible scientific breakthroughs rather than short-term products. He contrasts this with the current race dominated by chatbots from firms like OpenAI and Anthropic. These systems are described as “reversible,” easily replicated and surpassed. His approach prioritizes foundational discoveries over rapid commercialization cycles.

Evolvable AI raises new safety concerns

Evolvable AI (EAI) introduces systems capable of self-replication and adaptation in digital environments. Unlike traditional risks, these systems could spread like biological viruses without requiring high intelligence. Evolution occurs through copying models, prompts, and tools under selection pressures like compute and incentives. Experts warn this could become a fast-moving and hard-to-contain threat.

Hybrid Claude Code cuts AI costs

Claude Code workflows are being adapted to run with mixed local and cloud models, improving cost efficiency. Tools like VS Code, Ollama, and alternative models such as Nemotron and Mistral enable flexible setups. Users can balance privacy and performance by routing tasks across environments. This hybrid approach reflects a broader shift toward modular AI infrastructure.

Videos covered

Previous briefings · AI