AI news — March 10, 2026
The 16 top stories PulseAugur surfaced that day, ranked by signal across labs, papers, and developer communities.
-
IonRouter and RunAnywhere launch new AI inference and on-device solutions
IonRouter has launched a new inference stack called IonAttention, designed to multiplex models on a single GPU for high throughput and low cost, compatible with NVIDIA Grace Hopper. Separately, RunAnywhere has released RCLI, an on-device voice AI for macOS that runs inference lo…
-
Sequoia backs Scanner's AI-driven security logs search in seconds
Sequoia Capital has invested in Scanner, a company developing a novel log search engine designed for object storage. Their technology allows for petabytes of security data to be searched in seconds, addressing the high costs and limitations of traditional SIEM systems. This adva…
-
Microsoft Research unveils PlugMem to structure AI agent memory for better reuse
Microsoft Research has introduced PlugMem, a novel memory system designed to enhance AI agent performance by transforming raw interaction histories into structured, reusable knowledge. Unlike traditional methods that store text chunks, PlugMem focuses on storing facts and reusab…
-
Treasure hunter jailed for Ship of Gold coins released after 10 years
Tommy Thompson, a deep-sea treasure hunter famous for discovering the "Ship of Gold" (the S.S. Central America) in 1988, has been released from prison after serving over a decade. He was jailed for contempt of court after refusing to disclose the whereabouts of 500 missing gold …
-
Ash sandbox secures AI coding agents on Mac, limiting file and network access
Ash is a new macOS sandbox designed to enhance the security of AI coding agents like Claude. It restricts agents' access to sensitive system resources such as files, networks, and processes, mitigating risks of data exfiltration or accidental damage. Users define granular securi…
-
Didit launches Stripe-like identity verification platform with integrated AI
Didit, a startup founded by identical twins, has launched a new platform designed to streamline identity verification processes for businesses. The service aims to simplify the complex and fragmented landscape of KYC, AML, biometrics, and fraud prevention by offering a unified i…
-
Amazon mandates senior engineer sign-off on AI-assisted code changes
Amazon is implementing new protocols for AI-assisted code changes following a series of recent outages. Senior engineers will now be required to sign off on any modifications made with the help of generative AI tools. This decision comes after a significant site-wide outage that…
-
Open-source DD Photos generates fast, distraction-free photo albums
DD Photos is an open-source, self-hosted photo album generator built with Go and SvelteKit. It aims to provide a fast, distraction-free, and mobile-friendly way to share photos, addressing dissatisfaction with existing services like iCloud shared albums. The tool allows users to…
-
Mira Murati announces 1GW AI deployment with NVIDIA support
Mira Murati announced a partnership between an unspecified entity and NVIDIA to deploy 1GW of Vera Rubin systems. This collaboration aims to make adaptable, collaborative AI accessible to a wider audience. The initiative highlights the growing importance of specialized infrastru…
-
AI startup founded by Yann LeCun secures $1B seed funding
AI startup Mistral AI has secured a significant $1 billion in seed funding, marking the largest seed round ever raised in Europe. The funding round was led by Andreessen Horowitz and Lightspeed Venture Partners, with participation from other major investors including General Cat…
-
OpenAI enhances ChatGPT with interactive visuals for math and science learning
OpenAI has introduced interactive visual explanations within ChatGPT to enhance the learning of math and science concepts. This new feature allows users to manipulate variables and formulas in real-time, observing the immediate effects on graphs and outcomes. Initially covering …
-
AI code generation benchmarks may overestimate real-world usefulness
A new study from METR reveals that approximately half of the pull requests (PRs) generated by AI agents that pass automated SWE-bench tests would not be accepted by human maintainers. This discrepancy suggests that current benchmark scores may overestimate the real-world usefuln…
-
Yann LeCun launches AMI Labs with $1.03B seed for world model AI
Yann LeCun has launched Advanced Machine Intelligence (AMI Labs), a new startup focused on developing AI systems that understand the physical world. The company secured a substantial $1.03 billion in seed funding, reportedly at a $3.5 billion pre-money valuation, making it one o…
-
Anthropic faces developer ire as GitHub reliability plummets
Anthropic is facing developer backlash due to perceived model degradation and changes to Claude Code access, leading to speculation that capacity issues are the cause. Concurrently, GitHub's reliability has significantly worsened, with its leadership attributing the decline to a…
-
Hugging Face analyzes 16 open-source RL libraries for efficient training
Hugging Face's blog post details the landscape of 16 open-source reinforcement learning (RL) libraries, highlighting key lessons learned from their development and usage. The post emphasizes the importance of efficient token management and asynchronous training to optimize perfo…
-
Together AI enhances GPU clusters with multi-tenancy and autoscaling
Together AI has enhanced its GPU clusters with new features aimed at improving efficiency and manageability for AI-native teams. The platform now supports multi-tenancy, allowing different teams to share compute resources securely and independently. Key additions include autosca…