Latest in Artificial Intelligence
Curated from 3106+ articles across the top AI news sources. Updated hourly.
Subscribe to AI News
Get the latest AI news delivered to your inbox. No spam, unsubscribe anytime.
Male octopuses guided through mating by female hormones
Male octopuses use a receptor originally evolved for hunting prey to detect and respond to female mating pheromones, specifically progesterone. This discovery reveals an unexpected dual function of sensory receptors in octopuses' mating behavior. The finding suggests that evolution has repurposed existing sensory mechanisms to facilitate reproduction.
Neocloud Pioneer CoreWeave All In on Inference
CoreWeave, a GPU-as-a-service vendor, is shifting its business focus to specialize in AI inference rather than general GPU services. This strategic pivot represents the company's evolution as it targets the growing demand for inference workloads in the AI market.
New fossil deposits show complex animal groups predating the Cambrian
Recently discovered fossil deposits contain both Ediacaran and Cambrian species, providing evidence of a transitional period between these two major geological eras. The findings suggest that complex animal groups emerged and evolved more gradually than previously thought, rather than appearing suddenly during the Cambrian explosion. This collection offers new insights into early animal evolution and the diversification of life forms.
New Rowhammer attacks give complete control of machines running Nvidia GPUs
Researchers have discovered new Rowhammer attacks called GDDRHammer, GeForce, and GPUBreach that exploit GPU memory vulnerabilities to gain complete control over machines running Nvidia GPUs. These attacks can hijack the CPU by manipulating GPU memory, representing a significant security threat to systems relying on Nvidia processors.
KiloClaw targets shadow AI with autonomous agent governance
KiloClaw is a new governance tool designed to help enterprises manage and control autonomous AI agents, addressing the problem of "shadow AI" created when employees independently deploy AI solutions without oversight. The platform emerges as businesses realize that while they've been securing large language models and managing vendor agreements, workers have already begun deploying their own AI systems outside official channels.
Gemma 4: Byte for byte, the most capable open models
Gemma 4: Our most intelligent open models to date, purpose-built for advanced reasoning and agentic workflows.
Linear Regression Is Actually a Projection Problem (Part 2: From Projections to Predictions)
This article explores linear regression through the lens of vector projections, demonstrating how least squares regression can be understood as a geometric projection problem. The piece explains how this vector-based perspective provides intuition for how regression makes predictions by projecting data onto a lower-dimensional space.
Journalist Sues FAA Over Drone No Fly Zone Designed to Prevent Filming ICE
A Minnesota journalist is suing the FAA over a 3,000-foot drone no-fly zone established around Department of Homeland Security facilities, arguing the restriction violates First Amendment rights to film and report on government activities. The case centers on whether the FAA's airspace restrictions can be used to prevent media access and documentation of ICE operations. This raises broader questions about press freedom and government transparency in the drone age.
Artemis II Astronauts Have ‘Two Microsoft Outlooks’ and Neither Work
NASA's Artemis II astronauts encountered problems with Microsoft Outlook email software during their mission preparations, with two separate installations of the legacy application both malfunctioning. The issue highlights the challenges of using outdated software in critical space operations. This technical glitch underscores NASA's reliance on Microsoft products despite the limitations of older systems in demanding aerospace environments.
5 best practices to secure AI systems
As AI systems become increasingly powerful and integrated into critical operations, they introduce new security vulnerabilities that traditional security frameworks were not designed to handle. The article outlines five best practices for securing AI systems to address these emerging threats. Proper AI security measures are essential as the technology becomes more embedded in important business and infrastructure functions.
Microsoft to Invest $5.5 billion in AI in Singapore
Microsoft is investing $5.5 billion in artificial intelligence infrastructure and development in Singapore, continuing its global strategy of deploying sovereign AI capabilities across different regions. This investment underscores the tech giant's commitment to expanding AI services while respecting local data governance and sovereignty requirements.
A Secure Chat App’s Encryption Is So Bad It Is ‘Meaningless’
TeleGuard, a chat app with over a million downloads, claims to offer secure messaging but has critically flawed encryption that researchers describe as "meaningless." The app's fatal flaw is that it uploads users' private encryption keys to the company's servers, making message decryption trivial and defeating the purpose of end-to-end encryption.
How to Handle Classical Data in Quantum Models
The article explores workflows and encoding techniques for integrating classical data into quantum machine learning models. It addresses the fundamental challenge of converting traditional data formats into quantum-compatible representations that can be processed by quantum algorithms.
China’s Five-Year Plan details the targets for AI deployment
China's 15th Five-Year Plan through 2030 includes significant provisions for AI deployment across economic, educational, social, and industrial sectors. The plan positions AI as a key strategic technology alongside other emerging technologies to drive China's development goals.
Why wait until the end to realize your model’s code won’t actually run?
The article discusses the importance of testing and validating AI model code during development rather than waiting until the end when integration issues become apparent. It advocates for a "think anywhere in code generation" approach that enables developers to catch errors and verify functionality early in the development process. This proactive strategy helps prevent costly failures and ensures generated code will actually execute properly in production environments.
OpenAI acquires TBPN
OpenAI has acquired TBPN to accelerate global conversations about artificial intelligence and support independent media outlets. The acquisition aims to expand dialogue with developers, businesses, and the broader technology community.
Quantum Simulations with Python
This article discusses how to conduct quantum simulations using Python, specifically leveraging the Qiskit-Aer framework for running quantum experiments. Qiskit-Aer provides tools and simulators that enable developers and researchers to test and prototype quantum algorithms without requiring access to physical quantum hardware.
Codex now offers more flexible pricing for teams
Codex has introduced pay-as-you-go pricing options for ChatGPT Business and Enterprise customers, allowing teams greater flexibility in how they adopt and scale the service. This new pricing model makes it easier for organizations to start using Codex without committing to fixed plans upfront.
Autonomous AI systems depend on data governance
As AI systems become increasingly autonomous, data governance has emerged as a critical safety concern alongside model training and monitoring. Poor data quality—including fragmentation, outdated information, and lack of oversight—can directly compromise autonomous AI system behavior and reliability. The focus on AI safety is shifting from exclusively monitoring models to ensuring the integrity and governance of the data these systems depend on.
Experian uncovers fraud paradox in financial services’ AI adoption
Financial institutions are facing a "fraud paradox" where the same AI technology they deploy to fight fraud is being weaponized by criminals against them, according to Experian's 2026 Future of Fraud Forecast. Experian uniquely understands this tension as it operates on both sides—helping financial institutions detect fraud while also experiencing threats from AI-powered fraud tactics. This growing arms race between AI-driven fraud detection and AI-enabled fraud attacks presents a critical challenge for the financial services industry.
Want AI working for your business?
From AI receptionists to smart home automation — explore our full product line.