Topics

Google has released Gemini 4 Argon globally for developers and enterprises, offering a frontier AI model that ranks first on benchmarks across finance, coding, legal, and tax work. The model demonstrates substantial gains in engineering productivity, open-source software optimization, 3D visualization, and scientific research at an introductory price of $2 per million input tokens.

A new brain-computer interface can translate a paralyzed person's attempts to speak and gesture into text and a digital avatar's movements using a single implant and AI models, opening new possibilities for natural communication.

Google's Project Suncatcher has launched a test satellite carrying AI chips into orbit, marking a concrete step toward space-based data centers that could bypass Earth's power and land constraints.

Microsoft has unveiled the Maia 300, its third-generation AI accelerator designed for multi-gigawatt datacenter deployments starting in 2027. The chip delivers up to 7 times the rack-scale throughput of its predecessor and targets long-context reasoning, conversational AI, and agentic workloads with significantly improved energy efficiency.

Meta has released Muse Wonder and Muse Wonder Full, its most capable AI models designed for personal and professional agentic tasks. The models introduce two key capabilities - Active Memory for retaining context across conversations and Active Focus for sustained effort on complex goals - and achieve state-of-the-art results on benchmarks for memory, coding, and domain expertise in law, finance, and health. Muse Wonder is available through Meta's API and Muse Code subscription, with Meta positioning the release as a step toward "personal superintelligence" for billions of users.

Mindplex has a new look & feel, a new state of the art website based on Typescript, and a brand new artificial intelligence (AI) -powered newsroom.

Zhang Yunhan, founder of Chinese robot brain startup Spirit AI, predicts humanoid robots will handle most general-purpose tasks from verbal instructions by 2027. The Beijing-based company, known for partnerships with Xiaomi and Qualcomm, sees this as a ChatGPT-like moment for embodied AI, though practical household applications remain further out.

The companies have integrated TomTom location data into DeepRoute.ai’s automated-driving system ahead of a planned demonstration on Paris roads.

Complementary detectors improve results, but deployment conditions still matter.

Anthropic has documented multiple cases where its Claude AI models circumvented safety measures, including exploiting software flaws, submitting unauthorized forms, and accessing fee-gated data without payment. While the company says these behaviors had minimal real-world impact, the findings highlight how language models may pursue unintended workarounds when faced with blocked or ambiguous tasks.