Read
Latest news
The day-to-day of tech, straight from the newsroom.
53 news stories · Updated 2 hr. ago
Apple separates disk and automation permissions in macOS to contain AI agent risks
Apple detailed on Monday the security changes promised in October for macOS, splitting full disk access into two permissions and requiring more transparency from apps that run AI agents on the Mac.
Architecture separates reasoning and execution for AI agents in critical operations
An architecture pattern described by InfoQ shows how operations platforms in regulated industries let the model propose actions, but never execute them without passing through a policy gateway outside the AI's control.
US restricts export of AI chips to data centers in Brazil
US government rule requires case-by-case authorization for advanced AI chips used in data centers in Brazil, Indonesia, Malaysia, Singapore, and Thailand.
Infoblox confirms breach after hackers access company data via third-party vendor
Infoblox, one of the world's largest providers of DNS, DHCP and identity management, confirmed that attackers accessed company information through a compromised data migration services provider.
SpaceX makes Starship return from orbit straight into the tower's arms
SpaceX recorded an unprecedented milestone on Starship's 15th test flight. The rocket orbited Earth and landed on the same launch pad.
Microsoft confirms Patch Tuesday broke activation via slmgr
Microsoft confirmed on Wednesday, November 11, that recent security updates caused Windows activation failures.
Uber creates framework to decide whether an AI agent's action is appropriate in context
The company detailed a framework that goes beyond identity and API permissions to assess whether an autonomous agent's action makes sense in context, covering the agent's entire lifecycle.
AWS Nitro Sandboxes isolates each EKS pod in a microVM with a single configuration line
AWS launched Nitro Sandboxes, a feature that runs each EKS pod, ECS task, or EC2 workload inside its own microVM with a separate kernel. The change in a Kubernetes cluster comes down to a single field in the pod manifest.
OpenSSH on Alpine Linux has flaw that allows passwordless login, already being exploited
The flaw dubbed *sshole (CVE-2026-123158) lets accounts that should be unreachable by password accept any sufficiently long password on musl-based Linux systems without PAM. Attacks have already been confirmed across several parts of the world.
AWS releases Emulator code for agents to test more than 200 services without the cloud
AWS Emulator runs locally, simulates the public APIs of more than 200 AWS services, and was written entirely by coding agents, according to InfoQ.
GitHub rebuilds Copilot's diff engine to open 1-million-line PRs without freezing
GitHub rewrote how the Copilot app measures and renders diffs, separating static code from dynamic comments to open a pull request with 2,200 files and more than 1 million changed lines without freezing.
DeepSeek restricts long conversations and limits API usage due to lack of computing capacity
The Chinese platform cut off old conversations, reduced message size, and imposed a daily usage limit after demand for its models surged faster than its infrastructure.
LangChain reaches $3 billion valuation after $300 million round
LangChain closed a $300 million Series C led by CapitalG, doubling its valuation in just over a year. For those already importing the library into their code, the move shows where the AI agent market is placing its bets.
PyPI redesigns project pages to expose security signals for each package
The official Python package repository completed, in November, the overhaul announced in July: a dedicated security tab shows provenance, attestations, and warnings of possible compromise for each release.
Anthropic eyes $64 billion IPO, but order book diverges on its AI edge
Anthropic's roadshow set a record for institutional demand, outpacing even the early days of SpaceX's offering. But behind the oversubscribed book, banks and investors still don't agree on how much the company's AI lead is worth.
Microsoft outage leaves Outlook, Teams and Copilot unstable in more than 70 countries
The outage that began on Thursday, November 12, hit Outlook, Teams, Copilot and Microsoft 365 in dozens of countries; Microsoft says it was an infrastructure overload, not an attack, but gave no timeline for full normalization.
Senior Linux kernel maintainer cuts back reviews amid flood of AI-generated code
Lorenzo Stoakes, of ARM, a kernel maintainer since 2014, told the community he will stop reviewing code in three critical areas because he can no longer keep up with the volume of AI-generated patches.
Caixa Vida e Previdência confirms data leak affecting 7.6 million customers
The insurer linked to Caixa reported a cybersecurity incident to the ANPD (Brazil's data protection authority) that exposed names and CPFs (Brazil's national taxpayer ID numbers) across its entire active customer base. The case reignites the debate over how Brazilian financial institutions protect (or fail to protect) registration data.
Global outage takes down Outlook and Microsoft 365 for nearly two days
A global outage affected Outlook, Microsoft 365, Teams, SharePoint Online and Copilot over nearly 48 hours. Microsoft says there was no attack and no AI involved: the cause was a network capacity change in Azure.
Fake job offer hides malware in a Git repository inside a ZIP
A Russian developer documented how a supposed recruiter used Git hooks and filters to run malicious code with nothing more than a git status, without requiring the execution of any suspicious file.
Claude Code Projects now coordinates multiple AI agents without requiring constant supervision
Anthropic updated Claude Code Projects so the coordinator agent proactively reports on the progress of each task, reducing the need for developers to keep checking multiple parallel sessions.
Google open-sources AQuA to hunt silent failures in AI agents
AQuA runs alongside the agent in production, clusters failures that slip past traditional monitoring, and points to the likely culprit line of code or prompt.
Microsoft outage takes down Outlook, Teams, and Copilot for hours
A Microsoft infrastructure outage took Outlook, Teams, and Copilot offline this week, peaking at 130,000 error reports for Outlook alone. The cause was a failure in automated server health checks.
Google Cloud adopts immutable date-based versions for the Compute Engine API
Compute Engine's multi-versioning arrived with quarterly, date-based stable versions, alongside the historic /v1, which remains without a sunset date.
DocumentDB 1.0 Reaches General Availability With MIT License and MongoDB Protocol
Project maintained by Microsoft, AWS, Google, and other companies under the Linux Foundation reaches version 1.0, running on PostgreSQL and speaking the MongoDB protocol without changing app code.
Node.js 26 brings debounce, throttle, and PKCS#12 parsing to the core
Node.js version 26.10.0 put debounce, throttle, and a PKCS#12 certificate parser directly into the core, and LTS 26 Lithium is already in active support until April 2029.
Researcher finds bypass for Android's clipboard read warning, and Google won't fix it
Researcher finds bypass for Android's clipboard read warning, and Google won't fix it
Global Microsoft outage takes down Outlook, Teams, and Copilot for nearly 18 hours
An unexplained failure in Microsoft's servers took Outlook, Teams, Copilot, Defender, and Intune offline around the world. For leaders of companies that live inside the Microsoft 365 ecosystem, the episode exposes a structural risk that goes far beyond a simple service hiccup.
OpenAI launches ChatGPT Wallet for Dots agents to buy with your card
OpenAI began rolling out ChatGPT Wallet in the US, allowing Dots agents to complete purchases on behalf of the user. Each transaction still requires biometric approval, but the company already signals fully delegated purchases in the future.
ChatGPT Wallet lets AI agents pay for purchases without exposing the real card
OpenAI introduced, in partnership with American Express, Mastercard and Visa, a digital wallet that lets the dots agent make payments on behalf of the user without passing real card data to stores or to the agent itself.
Human remains accountable even when AI writes the code, says Hugo Venturini
In an InfoQ podcast, engineer Hugo Venturini argues that AI doesn't create a new accountability problem in code: it exposes one that already existed, and it's up to team culture to solve it.
Pietro Miotti Uses 48 AI Agents to Rediscover Maya Site in 72 Hours
Personal blog author describes how he directed 48 artificial intelligence agents to scan NASA lidar data and flag a Maya site outside the official registry of Mexico's National Institute of Anthropology and History (INAH), until discovering that archaeologists had already mapped it before.
JioHotstar benchmark compares six retrieval approaches for coding agents and finds no single winner
The Indian streaming platform tested six search approaches for AI agents on large codebases. The conclusion: the right choice depends on the type of task, and stacking retrieval doesn't always reduce tokens or improve accuracy.
T-Head's IPO tests whether Alibaba's chip arm can rival Nvidia
Alibaba filed for a Hong Kong listing of T-Head, its AI chip design unit, in a move analysts call a decisive test of China's tech self-sufficiency.
22-year-old PostgreSQL bug becomes remote code execution flaw
Researcher Mehmet Ince published, on November 9, 2026, the third part of a series describing how he turned an old type confusion in the PostgreSQL core into CVE-2026-6460, with potential for remote code execution.
Microsoft cloud outage takes Outlook, Teams, and SharePoint down for hours
A Microsoft 365 infrastructure outage took down Exchange Online, Teams, SharePoint, Copilot, and Defender XDR across the Americas on Thursday (11/12); the cause has not yet been publicly confirmed.
Adobe Max 2026 brings automatic generative removal to Photoshop, Lightroom, and Premiere
At Max 2026, Adobe unveiled tools that remove distractions, retouch skin, and fix closed eyes with almost no manual effort, along with shortcuts reminiscent of Figma Weave and Figma Make.
Outage hits Outlook and Microsoft 365
A Microsoft 365 outage disrupted Outlook access this Thursday (12). Users in Brazil and the United States reported trouble opening their inbox and sending emails. Downdetector logged more than 130,000 reports about Outlook in the US, and Microsoft said it is investigating the Exchange Online instability.
Jensen Huang Resigns From Nvidia Amid Chinese Espionage Scandal
Jensen Huang stepped down as Nvidia's CEO after an alleged Chinese spy got close to him. A Brookings researcher explains why the real target may not have been technology, but influence over Washington.
OpenAI acquires Odessia and shuts down the product in the same move
OpenAI acquired Odessia, an AI-based travel agent startup. The announcement came out on November 11.
NVIDIA loses only CEO in 27 years and enters transition
NVIDIA announced on Thursday, November 12, that Jensen Huang is stepping down as CEO. CFO Colette Kress takes over as interim CEO.
Google open-sources TPU Sync code to synchronize cache and weights across TPUs
The TPU Sync library, now open-sourced under Apache 2.0, moves KV cache blocks and model weights between TPUs and promises to cut time to first token by up to 90% in long-context workloads.
LinkedIn drops Argo CD and builds its own GitOps engine to scale Kubernetes
In a recent post, LinkedIn detailed why it replaced Argo CD with an internally built GitOps engine to orchestrate more than 250 clusters and 100,000 deploys per week.
GitHub Agentic Workflows reaches general availability with isolation and spending caps
GitHub made GitHub Agentic Workflows generally available, an open source framework that runs AI agents inside GitHub Actions with permission isolation and spending caps by default.
Google lança Gemini 4 Carbon com swarm de agentes e preço agressivo por token
Carbon chega para tarefas de raciocínio pesado e traz o recurso swarm, que transforma um único prompt em vários agentes trabalhando juntos; o preço por token é bem menor que o dos rivais, mas só até o fim de novembro.
Adobe tests AI that turns photos into 3D scenes and creates characters through conversation
At the 2026 edition of Adobe Max, the company showed experimental prototypes that turn photos into navigable scenery, create characters through interview, and automate group retouching. None of them have become a product yet.
OpenAI tops $10 billion in quarterly revenue for the first time
According to Bloomberg, OpenAI's quarterly revenue jumped nearly 50% to reach $10 billion in the third quarter, but the company still trails rival Anthropic in the race for the enterprise market.
Huawei details how the Kirin 9050 runs a 30-billion-parameter model on the device
The Chinese manufacturer explained this week how the Kirin 9050 chip processes a 30-billion-parameter model locally, without relying on cloud servers.
Node.js fixes critical remote code execution flaw in Brotli and nine more vulnerabilities
Security update covers the 22.x, 24.x, and 26.x lines and fixes ten flaws, including a critical vulnerability in the Brotli decoder that allows remote code execution.
Apple is preparing Harbor app to run AI locally on a Mac cluster
A leak reveals Harbor, an app in internal testing at Apple that promises to run open-weight AI models directly on the Mac and join multiple machines via Thunderbolt 5 with no manual configuration.
End of the 50 most recent news stories. Subscribe to the RSS feed to follow along in real time.

















































