50 articles found
A quiet revolution is underway in how academic institutions access frontier AI capabilities. Amazon's Nova Forge program and the Stanford-AWS partnership signal a fundamental shift: universities are no longer locked out of the computational resources that define cutting-edge AI research. This democratization could reshape both academic research and industry innovation in unexpected ways.
Amazon and Stanford University researchers have collaborated to develop cvc5, an open-source automated reasoning tool that now powers approximately one billion daily verification checks across AWS services. The tool uses satisfiability modulo theory (SMT) to detect logical errors in code and is deployed in AWS features like Automated Reasoning checks in Amazon Bedrock and Identity and Access Management Access Analyzer to enhance security and reliability for customers.
Amazon has announced the return of its Nova AI Challenge for the second year, with ten university teams competing to build AI agents capable of handling complex multi-step software development tasks while maintaining security and reliability. For the first time, participating teams will have access to Amazon Nova Forge, enabling them to customize frontier AI models with tools and computational resources previously unavailable to academic research programs, addressing historical barriers to large-scale AI research in universities.
This article is a curated roundup of the 10 most significant robotics developments from January 2026, spanning major announcements including Boston Dynamics' Atlas humanoid working with Google's DeepMind, Microsoft's Rho-alpha vision-language-action model for robots, Amazon's acquisition of Rightbot for logistics automation, and autonomous system debuts from companies like Doosan Bobcat and Neocis. The piece highlights the industry's shift from theoretical capabilities to real-world deployments across humanoids, industrial automation, AI models, and commercial applications.
Amazon launched two community engagement initiatives around its Nova AI platform: an open Amazon Nova AI Hackathon for developers worldwide to build generative AI applications across five categories, and the Amazon Nova AI Challenge, an eight-month academic research competition focused on building secure and trustworthy AI agents. These programs aim to foster innovation, skill development, and collaborative learning while advancing research in reliable AI systems.
Amazon discovered a high volume of child sexual abuse material (CSAM) in its AI training data and reported over 1 million instances to the National Center for Missing and Exploited Children in 2025, but has declined to disclose the sources of the problematic content. The company claims it sourced the material from third-party external sources used for AI training and states it cannot provide actionable details to law enforcement, raising concerns from child safety advocates about data provenance and safeguards in AI development.
Binghamton University researchers published a study in the Journal of Organizational Behavior examining human-robot collaboration in the workplace, finding that companies may gain greater competitive advantage through complementary human-robot integration rather than simple worker replacement. The study argues that collaborative approaches foster better employee commitment and organizational learning, while purely substitutive automation strategies can be easily replicated by competitors and may undermine efficiency gains.
Microsoft announced the Maia 200, its successor AI chip built on TSMC's 3nm process, which the company claims delivers 3x the performance of Amazon's Trainium and outperforms Google's TPU seventh generation. The chip is beginning rollout to Microsoft's data centers today and will be used to host OpenAI's GPT-5.2 model and power Microsoft's AI services, with early access to the SDK being offered to academics and developers.
Agility Robotics, Amazon Robotics, ASTM International, AWS, Brain Corp, GM, Harmonic Drive, maxon, PickNik Robotics, QNX, RealSense, Robust AI, Tesla, Toyota Research Institute, Universal Robots, and...
iRobot has emerged from Chapter 11 bankruptcy following Shenzhen Picea Robotics' acquisition, with the company restructured as a privately held entity. As part of the deal, iRobot established iRobot Safe Corp., a U.S.-based subsidiary with an independent board to protect consumer data and maintain separation from Chinese ownership, addressing regulatory and privacy concerns that had previously blocked Amazon's acquisition attempt.
Amazon has launched 'Health AI,' an AI-powered assistant for its One Medical telehealth platform that provides personalized health guidance, explains lab results, manages medications, and answers patient health questions based on medical records. The tool is designed to complement rather than replace healthcare providers, with Amazon implementing HIPAA-compliant privacy measures, joining competitors like Google and OpenAI in developing AI healthcare applications.
Wikipedia has signed commercial training data deals with major AI companies including Amazon, Microsoft, Meta, Perplexity, and Mistral AI through its Enterprise program, allowing these firms faster, high-volume access to its 65 million articles. The partnerships aim to help Wikipedia recoup costs from server strain caused by AI companies' data scraping while ensuring its content is used by reputable sources rather than less scrupulous ones.
The Wikimedia Foundation announced commercial API access deals with Microsoft, Meta, Amazon, Perplexity, and Mistral AI through its Wikimedia Enterprise program, allowing major AI companies to access Wikipedia's 65 million articles at higher speeds and volumes. The deals represent a significant shift in how AI companies source training data, with Wikipedia addressing rising infrastructure costs driven by large-scale scraping and bot traffic while maintaining free public access to its content.
NVIDIA kicked off the year at CES, where the crowd buzzed about the latest gaming announcements — including the native GeForce NOW app for Linux and Amazon Fire TV sticks, community-requested flight-c...
Amazon Web Services' AI Labs demonstrates that reinforcement learning can effectively customize general-purpose AI agents for specific domains and business use cases, such as customer service and coding assistants, achieving improved task success rates even with small training datasets. The research leverages existing environment simulators and verifiable reward functions to adapt agents without requiring extensive ML expertise or computational resources.
CES 2026 featured major robotics announcements including Mobileye's $900 million acquisition of humanoid robot developer Mentee Robotics, Oshkosh's launch of an autonomous refuse collection robot (HARR-E), and Amazon's acquisition of Rightbot Technologies for truck unloading automation. Industry leaders discussed the critical role of vision, perception, and AI in robotics advancement, with Starship Technologies highlighting its fleet of over 2,700 autonomous delivery robots that have completed 9 million real-world deliveries.
Infosys will integrate Topaz AI with Amazon Q and Bedrock to boost generative AI use for clients in key industries and workflows.
Amazon's AGI Lab is developing reinforcement learning "gyms" to train AI agents on the fundamental micro-interactions required for reliable autonomous task execution, such as scrolling, clicking, form completion, and error recovery across legacy systems. Rather than focusing on high-level tasks like booking vacations, researchers emphasize that agents must master hundreds of low-level atomic behaviors with deterministic reliability before complex real-world applications become feasible.
Announced at the CES trade show running this week in Las Vegas, NVIDIA is bringing more devices, more games and more ways to play to its GeForce NOW cloud gaming service. Powered by GeForce RTX 5080-c...
Amazon published a roundup of its most viewed research papers and technical reports from 2025, highlighting advances in multimodal foundation models (Nova Premier and Nova 2), unified speech-text processing (Nova Sonic), frontier AI model safety frameworks, and formally verified cloud-scale systems. The publications showcase Amazon's work in large language models, responsible AI practices, and advanced reasoning capabilities across multiple modalities.
iRobot's cofounder Colin Angle attributes the company's bankruptcy to the failure of its acquisition deal with Amazon, which fell through due to regulatory concerns. The collapse of this major transaction marked a significant turning point for the once-dominant robotics company, despite decades of market leadership with its Roomba vacuum cleaner.
Carnegie Mellon University's School of Computer Science faculty members have been recognized among 63 recipients of Amazon Research Awards, highlighting academic research support in computer science and AI fields. The awards support faculty research initiatives across various areas of computer science and artificial intelligence.
Amazon has expanded its AI-powered Dialogue Boost technology to Echo smart speakers and Fire TV devices, enabling on-device audio enhancement that clarifies dialogue while suppressing background music and sound effects across any streaming service. The technology uses deep neural networks with innovative sub-band processing and pseudo-labeling techniques to run efficiently on consumer hardware, making it particularly beneficial for the estimated 20% of the global population with hearing loss.
Amazon has launched Amazon Nova Forge, a new service that enables organizations to build customized foundation models by mixing their proprietary data with Amazon's training data at multiple stages of model development. This "open training" paradigm addresses the distribution mismatch problem where public LLMs struggle with organization-specific use cases, while avoiding "catastrophic forgetting" that typically occurs when fine-tuning closed-weight models.
Amazon is leveraging large language models (LLMs) to improve the quality and consistency of its product catalog at scale, moving beyond category-specific machine learning models. The company uses prompt tuning to adapt LLMs to understand product attributes, semantics, and vocabulary, enabling them to correct and update product listings across millions of items while maintaining appropriate granularity in product descriptions.
Amazon is addressing inefficiencies in AI reasoning models that waste computational resources by overthinking simple tasks, generating 7-10 times more tokens than necessary. The company is developing metacognitive AI systems that can autonomously assess query complexity and allocate reasoning effort appropriately, rather than applying deep reasoning indiscriminately across all queries.
Amazon announced 63 research award recipients from 41 universities across 8 countries, providing funding and AWS credits for projects in AI security, agentic AI, and machine learning infrastructure. The awards support academic researchers investigating topics including healthcare applications, cybersecurity, and AI agent development, with recipients gaining access to Amazon datasets and AWS AI/ML services.
Amazon has developed Autonomous Threat Analysis (ATA), a security system using agentic AI and adversarial multi-agent reinforcement learning to automatically detect and counter cyber threats at machine speed. The system uses red-team and blue-team AI agents that execute real commands on isolated test infrastructure to validate threat detection capabilities and generate improved security rules, reducing typical weeks-long security testing cycles down to hours.
Partnering with professional security researchers and the academic community to build safer, more robust AI.
Challenge pushes teams to demonstrate measurable gains in secure-coding performance while building AI agents that advance real-world utility and reliability at scale.
Initiative will fund over 100 doctoral students researching machine learning, computer vision, and natural-language processing at nine universities.
Amazon vice president and distinguished engineer Marc Brooker explains how agentic systems work under the hood — and how AWS’s new AgentCore framework implements their essential components.
The collaboration will advance research in generative AI, robotics, natural language processing and cloud computing while fostering innovation in foundational and emerging technologies.
The submission opens October 1 and closes on November 12, 2025
From reimagining storage to serverless computing, Aurora continues to push the boundaries of what's possible in database technology.
Trained on millions of hours of data from Amazon fulfillment centers and sortation centers, Amazon’s new DeepFleet models predict future traffic patterns for fleets of mobile robots.
Translating from natural to structured language, defining truth, and definitive reasoning remain topics of central concern in automated reasoning, but Amazon Web Services’ new Automated Reasoning chec...
University teams battle to harden and hack AI coding assistants in head-to-head tournament
Top eight university teams move on to head-to-head finals focused on AI security for code generation.
Awardees, who represent 46 universities in 10 countries, have access to Amazon public datasets, along with AWS AI/ML services and tools.
By combining surveys with ads targeted to metro and commuter rail lines, Amazon researchers identify the fraction of residents of different neighborhoods exposed to the ads and measure ad effectivenes...
How Amazon is helping transform plastics through innovation in materials, recycling technology, sortation, and more.
Unique end-of-arm tools with three-dimensional force sensors and innovative control algorithms enable robotic arms to “pick” items from and “stow” items in fabric storage pods.
New tool lets customers build, train, and deploy machine learning models using only natural language.
Inaugural global university competition focused on advancing secure, trusted AI-assisted software development.
Prototype is the first realization of a scalable, hardware-efficient quantum computing architecture based on bosonic quantum error correction.
From cloud databases and anomaly detection on graphs to recession prediction and Amazon's new Nova foundation models, these are the most viewed publications authored by Amazon scientists and collabora...
Awardees, who represent 10 universities, have access to Amazon public datasets, along with AWS AI/ML services and tools.
In a keynote address at the latest Amazon Machine Learning Conference, Amazon academic research consultant, Stanford professor, and recent Nobel laureate Guido Imbens offered insights on the estimatio...
While large language models and other foundation models are well represented, traditional Amazon interests such as bandit problems and new topics such as AI for automated reasoning also get their due.
