Stellantis Halts Level 3 Autonomous Driving Program, Citing High Costs and Low Demand
Stellantis has officially suspended the development and launch of its Level 3 autonomous driving system, known as AutoDrive. The decision to halt the self-driving technology program is attributed to significant development costs, technical complexity, and a perceived lack of consumer demand for the advanced feature. While the company stated the AV technology is ready for deployment, it has been put on hold with no immediate plans for a market launch. This move is seen as a strategic reset for the automaker, which will now increase its reliance on external suppliers and technology partners like aiMotive for future autonomous systems. The work already completed on the Level 3 system will be integrated into future platform developments, reflecting the broader challenges legacy automakers face in the competitive autonomous vehicle sector.
Sources:
Nvidia Launches Jetson Thor, a Powerful Computer for AI-Powered Robots
Nvidia has announced the general availability of its new robotics computer, the Jetson AGX Thor. Described as a “brain for robots,” this powerful new system-on-module (SoM) is designed to run the next generation of AI-powered and humanoid robots. The Jetson Thor robotics platform is built on Nvidia’s Blackwell GPU architecture and is capable of running complex generative AI and large language models (LLMs) directly on the device for powerful edge AI capabilities. This enables real-time reasoning and interaction with the physical world. Compared to its predecessor, Jetson Orin, the new SoM offers up to 7.5 times more AI compute performance. Early adopters of the technology include major players in the robotics space such as Amazon Robotics, Boston Dynamics, and Figure AI. The developer kit is now available, aimed at accelerating the development of autonomous machines across industries like manufacturing, logistics, and healthcare.
Sources:
Google Launches Gemini 2.5 Flash Image for Advanced AI Image Generation
Google has introduced Gemini 2.5 Flash Image, a new state-of-the-art generative AI model for image generation and editing. This latest version allows for blending multiple images, maintaining character consistency for storytelling, and making specific edits through natural language prompts. The model is now accessible to developers through the Gemini API and Google AI Studio, and to enterprise clients via Vertex AI. To encourage adoption, Google has partnered with OpenRouter.ai and fal.ai to make the model available to a wider developer community. All images created or modified with Gemini 2.5 Flash Image will feature an invisible SynthID digital watermark to identify them as AI-generated or edited, promoting responsible AI usage.
Cloudflare Introduces Zero Trust Tools for Secure Enterprise AI Adoption
Cloudflare has launched new cybersecurity features for its Cloudflare One Zero Trust platform, aimed at helping organizations securely manage the use of generative AI applications. The new capabilities provide businesses with tools to monitor AI usage, prevent data loss, and protect against ‘Shadow AI’ through a centralized platform. A key feature is the ‘Shadow AI Report,’ which gives security teams detailed insights into which AI applications are being used by employees, tackling a major Shadow IT risk. Additionally, ‘AI Prompt Protection’ acts as a data loss prevention (DLP) tool, allowing for policies to flag, warn, or block employees from submitting sensitive information, such as source code, into AI models. These tools are designed to enable employees to utilize AI for productivity without compromising corporate security and privacy standards.
Sources:
Kubernetes 1.34 Released with Enhanced Scheduling and Resource Management
The Kubernetes project has officially released Kubernetes 1.34, codenamed ‘Of Wind & Will (O’ WaW)’. This update to the leading container orchestration platform introduces 58 enhancements, focusing on scheduler visibility, life cycle observability, and resource management. Key features now stable include Dynamic Resource Allocation (DRA) for more flexible use of custom hardware like GPUs, and a mechanism to ensure replacement pods are created only after the previous pod has fully terminated. The release also brings a Linux node swap capability to prevent abrupt workload terminations when pods exhaust their memory. Now in beta is an asynchronous scheduler API calls feature, which aims to reduce scheduling bottlenecks and increase throughput. Alpha features include ‘NominatedNodeName’ for earlier insight into pod placement and a ‘Container stop signals’ feature for better termination visibility. Additionally, this version introduces KYAML, a new dialect of YAML designed specifically for Kubernetes, which is currently in alpha testing.
Sources:
GitLab Patches Critical Authentication Vulnerability (CVE-2025-2246)
GitLab has released versions 18.3.1, 18.2.5, and 18.1.5 for its Community Edition (CE) and Enterprise Edition (EE) to address a critical security vulnerability. It is strongly recommended that all self-managed installations be upgraded immediately. The primary security issue addressed, CVE-2025-2246, is a missing authentication vulnerability in the GraphQL endpoint. This critical flaw could have permitted an unauthenticated attacker to access sensitive manual CI/CD variables by querying the GraphQL API, posing a significant risk to DevSecOps pipelines. GitLab.com and GitLab Dedicated customers have already been patched and do not need to take any action.
Sources:
IBM and AMD Partner to Build Quantum-Centric Supercomputers
IBM and AMD announced a partnership on August 26th to create quantum-centric supercomputing architectures. This collaboration will integrate IBM’s quantum processors with AMD’s high-performance classical computing hardware, including CPUs, GPUs, and FPGAs. The goal is to develop hybrid quantum-classical systems capable of solving problems that are currently intractable for either quantum or classical computers alone. IBM aims for this partnership to help achieve its goal of a fault-tolerant quantum computer by 2029. The joint effort will leverage open-source tools like Qiskit to accelerate progress in the quantum computing field.
Sources:
Quantum Computing Investment Benefits Classical Systems, New Study Finds
A new study suggests that investments in quantum computing are beneficial even if fault-tolerant quantum machines do not emerge in the near future. The research, published in SSRN, found that advancements in quantum research often lead to “technological spillovers” that improve classical computing systems. An analysis of over 51,000 firms revealed that classical computing companies were more likely to invest in quantum technology. The findings challenge the idea that quantum R&D is an all-or-nothing endeavor, highlighting its immediate value across the tech industry.
Sources:
The Apache Software Foundation Releases 2025 Annual Report
The Apache Software Foundation (ASF) has released its annual report for the 2025 fiscal year, detailing a year of growth and strategic initiatives. The report highlights the ASF’s role in shaping EU regulations such as the Cyber Resilience Act and the AI Act. A new VP of Tooling role was created to improve developer tools and infrastructure for its open-source projects. The foundation also launched a new fundraising model to support investments in security and compliance with global standards. Five projects, including Apache Answer and Apache DataFusion, graduated to Top-Level Project (TLP) status. The ASF has also begun to evolve its logo and brand identity and launched a new newsletter called Plus One.
Sources: