GIGAZINE
PulseAugur coverage of GIGAZINE — every cluster mentioning GIGAZINE across labs, papers, and developer communities, ranked by signal.
-
AI models' Mona Lisa art compared; Grok's output deemed strange
A comparison of AI models drawing the Mona Lisa revealed that Grok produced uniquely strange results. The study evaluated the quality and cost-effectiveness of images generated by various AI models, including GPT-5.6 an…
-
Alibaba's Qwen AI website source code contains 'hentai' and other sensitive keywords
The official website source code for Alibaba Group's AI model Qwen has been found to contain a large number of sensitive keywords, including "hentai" and "nude." These terms, along with references to pornographic websit…
-
Gemini 3.5 Pro, Grok 4.5, and Fable 5 updates emerge amidst GPT-5.6 restrictions
New AI models are reportedly emerging, with Gemini 3.5 Pro slated for a July release, potentially to circumvent government regulations. Separately, Grok 4.5 is said to offer performance comparable to Claude Opus 4.8 at …
-
China Approves NVIDIA H200 AI Chip Imports for Domestic Firms
China has reportedly approved the import of NVIDIA's H200 AI chips for domestic AI companies. This move aims to support the development and deployment of advanced AI technologies within China. The approval signifies a p…
-
Grok 4.5 public release imminent; OfficeCLI offers AI control over Office suite
Elon Musk announced that Grok 4.5 will be publicly available tomorrow, describing it as comparable to Opus in capability but faster and cheaper. Separately, a tool called OfficeCLI has been released, allowing users to g…
-
AI ROI Outside Tech May Take Years, Not Months, Economists Warn
Economists suggest that the return on investment (ROI) for AI adoption, particularly outside the tech sector, may take significantly longer than anticipated. While stock valuations often price in future profit increases…
-
Microsoft cuts AI costs by prioritizing its own models over OpenAI, Anthropic
Microsoft is reportedly reducing its reliance on third-party AI models from companies like OpenAI and Anthropic to cut costs. The tech giant is increasingly deploying its own in-house AI models, referred to as "MAI" mod…
-
AI struggles with character ambiguity, leading to monotonous creations
A new study suggests that AI-generated characters in creative works tend to be monotonous. Researchers hypothesize that AI struggles with maintaining "ambiguity," a key element in developing complex and engaging charact…
-
Public Utility's "Custom AI" Revealed as Mix of Existing Models; Infineon Launches AI Data Center Component
A public utility's self-developed AI has been found to be a mere combination of two existing models, revealed through weight analysis. Separately, Infineon has introduced a new SiC JFET with normally off functionality, …
-
Microsoft launches Copilot Cowork, cheaper than Claude Cowork · 8 sources tracked
Microsoft has launched Copilot Cowork for general availability, positioning it as a more affordable enterprise solution compared to Claude Cowork. Copilot Cowork is reportedly 30-40% cheaper and offers enhanced support …
-
Meta plans temporary AI data centers; Vertiv uses NVIDIA digital twins
Meta is reportedly planning to establish data centers within temporary structures, a move detailed by GIGAZINE. Concurrently, ZDNET Japan reports that Vertiv is integrating NVIDIA's digital twin technology into its AI d…
-
Runway video AI integrates with ChatGPT and Claude
Runway, a video generation AI service, has launched Runway MCP, a new integration that allows its features to be used within other AI chat services like ChatGPT and Claude. This enables users to generate videos and edit…
-
Google removes claims about Chrome's on-device AI data handling
Google has removed claims that Chrome's on-device AI features do not send data to its servers. This retraction follows scrutiny and concerns about user privacy and data handling practices related to AI functionalities w…