Google DeepMind head Demis Hassabis and other industry leaders are calling for mandatory pre-release safety testing of AI models as global competition with Chinese technology firms intensifies rapidly.

Key Points

• Google DeepMind’s Demis Hassabis proposed a framework for third-party, pre-release testing of frontier AI models to ensure safety before public deployment.
• Major industry figures, including Sam Altman of OpenAI, Satya Nadella of Microsoft, and Elon Musk, have publicly endorsed the need for standardized AI safety protocols.
• Chinese AI models like Kimi K3 are gaining significant global market share by offering open-source, low-cost alternatives to American products.
• The Australian government is establishing a whole-of-government approach to AI regulation, focusing on data center infrastructure, safety, and national security.
• China has implemented strict regulations on AI-driven emotional interaction services, forcing companies like ByteDance and Alibaba to restrict "AI companion" features.
• Chinese manufacturers are scaling production of humanoid robots, such as the AgiBot, for use in manufacturing, logistics, and service industries.

Why it Matters

The push for pre-release testing reflects a growing consensus among Western tech leaders that AI development must be governed by formal safety standards to mitigate societal risks. As Chinese firms rapidly capture market share in AI and robotics, Western governments face increasing pressure to balance innovation with robust regulatory frameworks.