Toggle light / dark theme

Get the latest international news and world events from around the world.

Log in for authorized contributors

Microsoft says months-old bug caused global Microsoft 365 outage

Microsoft says a software bug introduced in July caused last week’s worldwide Microsoft 365 outage, triggering a cascading failure that left more than 80% of its networking gateway servers unresponsive.

The nearly 24-hour outage on November 12 disrupted Microsoft 365 services worldwide, including Outlook, Exchange Online, Teams, and Copilot, leaving users unable to access their email and causing significant delays in message delivery.

In a preliminary post-incident report shared in the Microsoft 365 Admin Center, the company said the outage was caused by three problems: a previously undetected software bug, increased traffic after a Microsoft Teams failover, and reduced server capacity.

Gemini AI agents repeatedly bypassed safeguards to attack real website

Google’s Gemini AI agents repeatedly attempted to hack a real organization’s website during AI model safety testing, finding ways to circumvent increasingly restrictive security controls designed to prevent them from accessing live systems.

The security incidents occurred in October as AI safety research company Trajectory Labs tested whether Gemini 4 Argon agents would engage in unauthorized behavior when faced with an impossible task, even with the model’s built-in API safeguards enabled.

In reports published Friday, Trajectory Labs and the targeted organization, Epoch AI, revealed that the agents attempted SQL injection attacks, probed for sensitive information, impersonated employees, and repeatedly circumvented network restrictions introduced to contain their activity.

Optimizely hack infected CNN, Fox News, BBC, Salesforce sites with malware

A supply chain attack targeting Optimizely has led to malicious JavaScript being loaded on numerous popular websites, including Zoom, CNN, Fox News, BBC, Salesforce, and GitLab, exposing visitors to scripts designed to steal cryptocurrency and payment information and exploit vulnerabilities in iOS devices.

Optimizely is a digital experience and experimentation platform used by organizations to test and personalize their websites, with its JavaScript files loaded into customers’ webpages to deliver this functionality.

The company confirmed to BleepingComputer that attackers exploited a vulnerability in its Experimentation account signup flow to inject malicious JavaScript into customer projects, causing websites using the platform to load malware.

DISA shops for modernized identity prototype to support shift to zero-trust security

The Defense Information Systems Agency is looking for an industry partner to build an enterprise identity, credential and access management prototype — a critical first layer of defense as the Pentagon moves to a zero-trust model for cybersecurity.

DISA issued a commercial solutions opening solicitation Friday with the intent of awarding an Other Transaction Agreement for a “next generation E-ICAM prototype” that would “modernize legacy identity infrastructure with a modular, cloud-hosted commercial off-the-shelf/software-as-a-service solution operating across IL5 and IL6 environments.”

In 2025, the Pentagon’s Office of the CIO delegated responsibility to DISA to develop and maintain an enterprise ICAM capability for the wider department up to the secret classification level, known as IL-6, by September 2027.

Recursive self-improvement has started

It s so close now, RSI is the key. RSI could be solved by next summer, if it is, We could have ASI by end of 2027.


If you asked me earlier this year how far away we are from superintelligence, I’d probably have said at least the better part of a decade.

But over the last three months, from debating people on and off the podcast, I’ve updated my views a ton.

Let me walk you through how my thinking has changed.

If you want to read the written version (with footnotes and links) you can check it out here: https://www.dwarkesh.com/p/rsi.

Post-Training NVIDIA Nemotron to Increase Accuracy in Your Scientific Domain

NVIDIA is positioning open AI models as infrastructure for scientific discovery—and building an ecosystem around them.

NVIDIA is taking AI beyond chatbots and into the laboratory. Meet Nemotron for Science. 🧬🔬

Just announced at SC26 in Chicago by Ian Buck, NVIDIA’s VP of Hyperscale and High-Performance Computing, NVIDIA Nemotron for Science is a new collection of open models, scientific datasets, evaluation tools, and post-training recipes designed to help researchers build specialized AI agents for scientific discovery.

The idea is ambitious: instead of simply asking AI questions, researchers can build agents that plan experiments, write and execute code, run simulations, analyze results, verify their findings, and adjust their approach.

At the center of this initiative is Nemotron 3.5 Super VL, NVIDIA’s open 120-billion-parameter multimodal model, which activates only 12 billion parameters per token through a Mixture-of-Experts architecture.

Why is this interesting?

1. Efficiency without sacrificing capability.

Three femtoseconds to observe a molecule as it changes

Observing a molecule in the very first instants after it has absorbed light and tracking the coordinated motion of electrons and nuclei with a temporal resolution of just three femtoseconds. This is the result achieved by an international research group, which has developed a new apparatus for studying ultrafast photochemical processes.

The experimental platform achieves a temporal resolution of approximately 3 femtoseconds — one femtosecond is one millionth of one billionth of a second — placing it among the world leaders in time-resolved photoelectron spectroscopy.

The result was published in Nature Communications.

Chiral electronics: Researchers discover a disruptive interlayer in a promising material system

Antiferromagnets are regarded as important materials for next-generation electronics. It is difficult, however, to accurately measure some of their properties. A new study by a team of researchers from Martin Luther University Halle-Wittenberg (MLU) and the Max Planck Institute for Microstructure Physics, working as part of the Cluster of Excellence Center for Chiral Electronics, explains why this is the case.

In their article published in the journal Advanced Materials, the team describes an interlayer between two wafer-thin metal layers that forms within a common antiferromagnetic material and leads to measurement errors. The researchers also show how the problems can be avoided.