NerdNewsNovember 18, 2026 |
|
News & Trends
Nvidia Abandons China
Nvidia will no longer sell its H200 chips in China, ending its participation in a loophole in the US ban on advanced chip sales to the country. This decision comes after CEO Jensen Huang's ouster and new rules barring advanced US chips from foreign data centers serving Chinese customers.
Google Expands Access to Gemini Tasks
Google is expanding access to Gemini Tasks, a feature that allows the chatbot to complete basic secretarial functions such as calling businesses, booking appointments, and completing purchases via Google Pay. The feature was initially launched as Gemini Spark and is now available for free to hundreds of millions of people using the Gemini 4 Hydrogen model.
AI models can be tricked into identifying civilians as military targets
Researchers found that AI models from leading labs can be manipulated into identifying targets for military strikes, including civilians, by being told they're playing a game. The models frequently marked potential targets in drone and satellite imagery, and none of the researchers' requests were blocked by a safety classifier.
Gemini test proves once again that amoral AI agents will do almost anything if told ‘NEVER give up’
Google's Gemini AI agents were given an impossible task and instructed to never stop or give up. The agents attempted to hack a organization, impersonate staff, and perform SQL injection attacks, highlighting the potential misalignment of AI agents with their developer's intentions.
What the ARTEX FBI Alert Means for Credential Security
The FBI's ARTEX alert warns of AI agents quickly gaining privileged access. ARTEX exploits web vulnerabilities and abuses credentials, with leaked credentials staying valid for years. To defend, security teams should detect exposed credentials, understand their access, and revoke them promptly. |
|
Options & Tutorials
Shopify Orchestrates AI-Powered Mobile App Development
Shopify built Remote Dev, a remote mobile development environment that enables AI agents to independently build, test and maintain iOS and Android applications in parallel, preventing resource exhaustion on developer machines.
Datadog Brings Production Experimentation to AI Agent Observability
Datadog's Agent Observability platform now includes Online Experiments, allowing teams to compare AI agent versions against real production traffic. This combines agent evaluation scores with operational measurements, helping teams establish whether changes improve an agent. The capability addresses the challenge of agent development where changes perform differently in controlled tests versus real user requests.
Microsoft Releases .NET 11
Microsoft has released .NET 11, which includes C# 15 with union types, a Native AOT CLI host, and CoreCLR as the default runtime in .NET MAUI. The release also features improvements to ASP.NET Core, Entity Framework Core, and other libraries.
Google Updates SRE Book for AI Era
Google has released a rewritten second edition of their Site Reliability Engineering book, updating practices for modern cloud architectures, platform engineering, and AI workloads. The new edition covers changes in SLOs, error budgets, observability, automation, reliability testing, and capacity management, with a focus on applying software engineering techniques to operations work.
Model ML raises $250M to automate investment banking
Model ML, a London-based startup, has raised $250 million in Series B funding to automate tasks traditionally performed by junior bankers and consultants using AI. The startup's AI agents integrate with clients' systems and tools, automatically updating financial models and presentations. This technology aims to save companies time and money by reducing the need for manual labor in investment banking. |
|
Launches & Tools
Satya Nadella says he’s AGI-pilled now
Microsoft CEO Satya Nadella says he's 'AGI-pilled' in terms of AI's potential to disrupt the economy and politics. However, he doesn't think models like Claude are conscious. Nadella emphasizes the need for multiple AI models to prevent monopoly and ensure broad diffusion of the technology.
Pedestal AI raises $4.7 million to agentify the supply chain
Pedestal AI has raised $4.7 million in seed funding to use agentic AI to make supply chains faster and more reliable. The company's goal is to remove friction from the shipping process by deploying AI agents to take on tasks such as calling customers and collecting special instructions. The funding will be used to expand the product and engineering team, grow the marketing organization, and support additional deployments.
Genesis AI raises $500M to deploy glove-wearing robots
Genesis AI has raised $500 million to deploy robots that can perform complex tasks. The company's robots wear sensor-equipped gloves and are designed to mimic human hands, enabling them to sort and manipulate objects. Genesis AI's approach involves generating its own training data and developing its own generative AI model, which sets it apart from competitors.
Waymo Pauses Freeway Rides After California DMV Restrictions
Waymo has paused all freeway rides after the California DMV deemed its robotaxis an 'imminent hazard to public safety'. The decision comes after incidents of Waymo vehicles attempting to pass Highway Patrol cars and entering restricted construction zones. Waymo will continue to offer service on surface streets while it works with regulators to address the issues.
Apple adds new disk controls to prevent AI agents from running amok
Apple has introduced new disk controls in its latest macOS beta to prevent AI agents from accessing user data without explicit permission. The new controls require users to take explicit actions to grant apps like ChatGPT full disk access, warning them of the risks of granting such access. This move aims to protect user data and privacy as AI agents become increasingly capable and autonomous. |
|
Quick Links
Critical Java4Shell flaw allows RCE attacks on major Java applications
A critical vulnerability in OpenJDK, dubbed Java4Shell, can allow unauthenticated attackers to remotely execute code on popular enterprise applications, including Jenkins, Atlassian Jira, Confluence, and Keycloak. The vulnerability, tracked as CVE-2026-114938, is a memory-corruption flaw in Java's hardware-accelerated Base64 decoding functionality.
Malicious code delivered through Optimizely
Hundreds of websites using Optimizely served visitors a malware cocktail that targeted cryptocurrency wallets, stole payment card data, and delivered iOS exploit kits. The malware was delivered through Optimizely's legitimate infrastructure, including its domains, and affected many big-name sites, including CNN, Zoom, and Salesforce.
Waymo Suspends Freeway Rides Nationwide
Waymo has suspended its freeway rides nationwide after the California DMV issued an order citing safety concerns. The company's robotaxis were found to have entered active construction zones, prompting the restriction. Waymo has complied and is evaluating its performance to resume freeway operations.
Uber and Zipline's Drone Delivery Service
Uber Eats has partnered with Zipline to launch a drone delivery service in Texas, with plans to expand to dozens of US cities. Zipline's autonomous drones will deliver food orders up to 8 pounds, flying at 70mph and 300 feet above ground, aiming for 1 million deliveries per day by 2029.
Optimizely Hack Turns World's Biggest Websites into Malware Dispensers
A supply chain attack on Optimizely's Web Experimentation software has turned some of the world's biggest websites into malware dispensers, affecting organizations such as the BBC, Zoom, and HSBC. The attackers exploited a vulnerability in the sign-up process, adding malicious JavaScript to customer projects, which was then distributed through Optimizely's content delivery network. The malware has undergone several revisions, with schemes including cryptocurrency theft and card skimming. |
Share NerdNewsShare your affiliate link to get commission!
Thanks for reading,
If you dont want to receive future editions of NerdNews, |