NadiAIGlobal AI news. Concise. Clear. Easy to follow.
Meta published Glimmer, an open-weight model users can download and run on their own hardware, while keeping a more powerful model, Muse Spark, behind APIs. The company accompanied the release with a letter from Mark Zuckerberg saying AI should be "for everyone."
OpenAI is expanding its Daybreak AI cybersecurity defense program and rolling out a new cyber-trained model alongside it, the company says. TechCrunch reports the move is positioned to address a rise in AI-led attacks.
Arcee, a US open-source AI lab, argued that Chinese AI models are not inherently dangerous as their capability and use grow among U.S. companies. The comment comes amid intensified debate in the U.S. over how to respond to those models.
Google DeepMind and Isomorphic Labs describe their joint approach to bioresilience and AI model development. The post outlines principles and collaborations intended to strengthen biological risk understanding using AI tools.
TechCrunch reports Microsoft is coaching its sales team to present Microsoft’s own AI models as more efficient and cost-effective than rivals like OpenAI and Anthropic. The guidance appears aimed at winning enterprise customers by emphasizing operational and pricing advantages.
Several U.S. companies are opting for faster, lower-cost Chinese AI models instead of pricier American alternatives, according to reporting. Businesses say the cheaper models meet practical needs even if they are not the most advanced available.
Semasa negara meraikan kemenangan Piala Dunia dan Knicks, Anthropic menghabiskan hujung minggu bergelut dengan pentadbiran Trump mengenai pelancaran model terbarunya. Syarikat itu menerima arahan kawalan eksport AS pada Jumaat jam 17:21 untuk menggantung akses kepada Mythos 5 dan Fable 5.
OpenAI mengumumkan GPT OSS, keluarga model sumber terbuka baru. Menurut Hugging Face Blog, pengumuman ini berfokus pada penggunaan dan kolaborasi komuniti pembangun.
