Full Summary
This Thursday morning, multiple sources confirm a significant surge in the AI landscape, with Zhipu AI's new model driving its market value past one trillion Hong Kong dollars. Both iChongqing and RS Web Solutions report this Chinese company's stock climbed following the launch of GLM-5.2, a flagship language model with a one million-token context window, excelling in complex coding and enterprise tasks. GLM-5.2 even topped open-source model rankings on Artificial Analysis and achieved the highest scores for an open-weight model on ARC-AGI benchmarks, according to OfficeChai. Meanwhile, the US and India are engaged in high-level, sensitive discussions regarding the rollout and security of advanced AI models, including Anthropic's Fable. The Times of India, EdexLive, and Moneycontrol.com all highlight US Under Secretary Jacob Helberg's confirmation of these talks, emphasizing a "gradual, measured approach" for safety, especially concerning critical infrastructure. India, through S. Krishnan, has expressed concerns about potential disruptions in access, seeking assurances against sudden cutoffs. What nobody expected is the Trump administration's intervention in OpenAI's GPT-5.6 launch, delaying its full release due to security concerns, as reported by The Tech Buzz. The model will only be available in a "limited preview," with federal officials personally approving each enterprise customer. This marks an unprecedented level of government involvement in AI deployment. In other developments, Google has postponed the launch of Gemini 3.5 Pro to July 2026, delaying it to refine performance and improve autonomous AI agents, according to Lapaas Voice. Domyn, an Italian company, plans to launch a fully open-source frontier AI model within a year, aiming to build one of the most advanced systems with over 400 billion parameters, a move noted by both The Economic Times and Crypto Briefing. This initiative, part of Europe's push for AI sovereignty, will allow companies and governments to run the model on their own infrastructure. For your daily life, these developments mean a tightening grip on how powerful AI models are released and accessed, potentially affecting the features and availability of AI tools you use at work or in personal applications.