Tech & AI News: Quantum Leaps, Autonomous Driving, and Generative AI Updates
GM Unveils ‘Eyes-Off’ Autonomous Driving for 2028 Cadillac Escalade IQ
General Motors has announced its next-generation autonomous driving technology, enabling hands-free and eyes-free operation on highways. This advanced system is set to debut on the 2028 Cadillac Escalade IQ electric SUV. The system will leverage a combination of lidar, radar, and cameras to build a comprehensive understanding of the vehicle’s environment.
In addition to this leap in autonomy, GM is integrating Google’s Gemini for conversational AI in its vehicles starting next year, with plans to introduce its own proprietary AI fine-tuned for the vehicle and connected to OnStar. The 2028 Escalade IQ will also feature a new centralized computing architecture, unifying all major vehicle systems onto a single computer core to significantly boost bandwidth and over-the-air update capabilities. This announcement signals a renewed focus on autonomous vehicle technology for personal use, less than a year after the company shut down its robotaxi unit, Cruise.
Sources:
Avride Secures $375M from Uber to Scale Autonomous Robotaxi Fleet
Avride, a developer of autonomous robotaxis and delivery robots, has secured up to $375 million in strategic investments and commercial commitments from Uber and Nebius. This funding will facilitate the rapid expansion of their autonomous fleet to as many as 500 vehicles and advance their AI-driven product development and geographic reach.
A spin-off from the Yandex self-driving car group, Avride was founded in 2020. The company is focusing on both passenger transport and delivery robots, a strategy that differentiates it from competitors. Avride plans to launch its public robotaxi service on Uber’s platform in Dallas before the end of 2025.
Adobe Launches AI Foundry for Custom Enterprise Generative Models
Adobe has introduced AI Foundry, an enterprise platform that enables businesses to train custom generative AI models using their own proprietary assets. This service allows companies to create brand-specific models for generating video, 3D designs, and text that align with their unique brand identity and tone.
The platform is designed to ensure all AI-generated content remains consistent with company branding while protecting intellectual property. Adobe is shifting towards a usage-based pricing model for this service. Early adopters of the AI Foundry include Home Depot and Walt Disney Imagineering, who are using it to automate and scale their content creation and campaign workflows.
Sources:
OpenAI Releases Sora 2 Text-to-Video Model with Commercial Licensing on Azure
OpenAI has launched Sora 2, its next-generation text-to-video model, which is now available in public preview on Microsoft’s Azure AI Foundry. This marks the first time the model’s API is accessible via a cloud platform, signaling a move towards the commercial use of generative AI video tools.
Sora 2 is designed to generate high-quality, cinema-grade video content from text prompts, featuring improved temporal coherence and the ability to create longer clips. The model is aimed at video creators, marketers, and filmmakers looking to streamline their production processes. Additionally, platform updates now allow all users to generate videos up to 15 seconds, with Pro subscribers having an extended limit of 25 seconds on the web.
Sources:
Anthropic Launches Claude Haiku 4.5 for Real-Time AI Applications
Anthropic has released Claude Haiku 4.5, a new compact AI model designed for real-time, low-latency applications such as chat assistants and customer service. This model is now available to subscription users on Claude’s web and app platforms at no additional cost. For developers, the pricing is set at $1 per million input tokens and $5 per million output tokens.
The release of this more affordable and high-quality model aims to make advanced AI workflows more accessible to smaller teams. In a related development, Anthropic has also introduced a “Skills” feature for its AI assistant, Claude, which allows users to load specific instructions and tools to automate tasks and improve workflow efficiency.
Sources:
Google Launches Gemini Enterprise and Upgrades Veo Video Model
Google has launched Gemini Enterprise, a conversational AI platform for corporate clients that allows employees to interact with internal company data, documents, and applications through a chat interface. The platform includes pre-built agents for analysis and tools for creating custom AI assistants, with early adopters including Gap, Figma, and Klarna.
In other news, Google has also released Veo 3.1, an upgraded version of its video generation model. This new version adds synchronized audio generation to all video outputs, creating a more immersive and complete experience.
Sources:
SAP Enhances Business AI with Joule Studio and New Model Integrations
SAP has announced its Q3 2025 release highlights for Business AI, featuring the new Joule Studio agent builder, which allows users to create custom AI agents for automating complex business processes. The generative AI hub in SAP AI Core has been updated to include new models such as GPT-5, Claude Sonnet 4, and others from AWS Bedrock and Cohere.
This provides developers with access to over 40 AI models through a single endpoint. Additionally, SAP has embedded generative AI assistance within SAP Build Work Zone to help with content creation and summarization. New AI features have also been added to SAP Analytics Cloud, enabling users to generate and understand complex calculation formulas using natural language.
Sources:
Google’s ‘Quantum Echoes’ Algorithm Achieves Breakthrough Performance
Google’s Quantum AI division has announced a significant breakthrough with a new algorithm named ‘Quantum Echoes’. In a study published in the journal Nature, researchers detailed how the algorithm, running on Google’s Willow quantum processor, performed a complex physics simulation 13,000 times faster than the Frontier supercomputer, the world’s leading classical machine.
This marks the first instance of a quantum computer executing a verifiable algorithm that demonstrably outperforms a supercomputer. A key aspect of this achievement is that the results can be independently verified, addressing a critical challenge in the field. The experiment, which would have taken the Frontier supercomputer over three years, was completed in just over two hours on the quantum device. This advancement is a major step toward practical applications in fields like drug discovery and materials science.
Sources:
Qilimanjaro Releases QiliSDK for Hybrid Quantum Computing
Qilimanjaro Quantum Tech has released QiliSDK, an open-source Python framework for designing digital, analog, and hybrid quantum algorithms. This software development kit (SDK) is part of Qilimanjaro’s effort to provide a full-stack toolkit for practical quantum computing, aiming to accelerate the transition from theory to hardware implementation.
The QiliSDK provides an entry point to Qilimanjaro’s multimodal quantum data center, which integrates superconducting digital and analog QPUs with classical HPC accelerators. It allows researchers and developers to run and optimize quantum workflows on various backends, from CPUs and GPUs to Qilimanjaro’s superconducting QPUs.
Sources: