Jeff Jarvis, a US media critic, warns that AI companies are buying the silence of publishers. He criticizes the exclusive partnerships between AI companies and media corporations. According to him, tech companies are silencing publishers in legal disputes. While companies like Meta and OpenAI make deals with large media houses, smaller and local publications are left out.
Jarvis argues that AI training with media content should fall under the “Fair Use” rule. He suggests media companies develop new business models with AI, such as a joint news API. Given AI’s ability to create massive content, he advises journalists to adopt “community journalism,” serving as service providers for specific communities. Jarvis is particularly critical of leading AI entrepreneurs like Sam Altman and Elon Musk, whom he calls “prophets of doom.” He warns against their ideologies like transhumanism and their claim to set the rules for humanity’s future.
Elon Musk’s AI start-up xAI plans to launch a chatbot app in December to compete directly with established offerings like ChatGPT. Currently, xAI’s chatbot Grok is only available to subscribers of Musk’s social media platform X. To catch up, xAI has heavily invested in infrastructure, including a data center with 100,000 Nvidia GPUs. The start-up is valued at $50 billion but lags behind in revenue compared to competitors. Technically, xAI’s language model Grok-2 is on a similar level to GPT-4 and Claude 3 Opus but only draws new information from X postings, not from the internet. Given the issues with right-wing propaganda and misinformation on the platform, the value for users is questionable.
Red Hat’s OpenShift AI platform receives numerous new features in version 2.15, including a Model Registry for centralized model management. This allows the sharing, versioning, deployment, and control of predictive and generative AI models, metadata, and model artifacts. OpenShift AI is a platform for developing and operating AI-powered applications in a hybrid cloud environment. Its tasks include data collection and preparation, model training, fine-tuning, model deployment, monitoring, and hardware acceleration. Red Hat aims to be an application-neutral platform provider, allowing platform-independent operation of AI models.
However, there is an increasing problem: large language models are strongly tied to the hardware they run on. To solve this, Red Hat recently acquired Neural Magic, an MIT spin-off, and plans to launch an “Open Hybrid AI.” This requires abstracting LLMs so they can run on any platform, called “virtual LLMs” (vLLMs). The concept was developed at the University of California, Berkeley, and aims to allow AI models on various hardware platforms, including processors from AMD, Intel, Nvidia, and custom chips from Amazon Web Services and Google.
The Gemini app for the AI assistant is now available for Google Workspace. Users can use the chat as a research tool and the camera to transfer handwritten notes to Google Docs or Gmail or convert drawn diagrams into digital graphics. Privacy settings from Google Workspace apply. Users can download Gemini for Android and iOS from the respective stores. However, not all web version features are available in the app. Files cannot be sent to the AI assistant, and Gemini’s themed chatbots and Android work profiles are not yet supported.
Hamburg-Holstein transport companies plan to deploy autonomous shuttles in the Harburg district from the second half of 2025. The first 6.9-meter-long electric vehicle from Austrian manufacturer eVersum offers nine seats plus space for a wheelchair or stroller. Four more shuttles are to follow, with a fleet of up to 20 vehicles planned. The Federal Ministry for Digital and Transport supports the project with 18 million euros. Operations will start with a test phase with safety drivers but without passengers. In a third phase, passengers will be transported. The vehicles should be able to operate autonomously at Level 4, meaning they can handle all situations independently in defined scenarios. The project runs until June 2027.
The AI start-up Cradle for protein engineering has raised $73 million in a financing round, increasing total funding to over $100 million. Cradle’s AI platform allows researchers to accelerate and reduce the cost of discovering and developing improved proteins. The funding is intended to expand Cradle’s wet lab and improve machine learning capabilities. The company, based in Amsterdam and Zurich, has partnerships with companies like Novo Nordisk and Ginkgo Bioworks.
The AI start-up /dev/agents has raised $56 million to develop an operating system for AI agents. The goal is for computers to collaborate like humans. This requires new user interfaces, a revised privacy model, and a simplified developer platform. The founding team includes experienced tech veterans who previously worked at companies like Stripe, Google, Facebook, Meta, Dropbox, and Figma.