AI News, 15 November 2026

Sunday 15 November

How do OpenAI, Microsoft, Anthropic and Google AI bug hunters compare in November 2026?

Image Microsoft published with the announcement covered in this story

AI-generated summary Four labs published research describing how OpenAI, Microsoft, Anthropic and Google AI models found vulnerabilities in real software. OpenAI validated 948 Linux kernel flaws, Anthropic chained bugs into Chrome and Linux exploits, Microsoft turned findings into kernel reports, and Google DeepMind weakened 1024-bit RSA.

At least one U.S. state passes legislation restricting new AI data-center construction before the end of 2027. — 52% say yes

AI-generated summary Midterm election outcomes have increased scrutiny over power and land use by AI facilities across the nation. Frontier models estimate a median probability of 52 percent that at least one U.S. state passes legislation restricting new data-center construction before the end of 2027.

Earlier:

Meet the short seller that flagged Nvidia's 'significant China problem' six months ago—including ‘uncomfortably close’ ties with Alice Huang

Jensen Huang, co-founder and CEO of Nvidia, speaks during GTC Berlin. The U.S. chipmaker held its AI conference.

AI-generated summary Culper Research raised concerns about Nvidia regarding its exposure to China and ties with Alice Huang six months ago. The short seller alleged that Nvidia obtained significant revenue from Chinese demand despite export restrictions.

Earlier:

Deep Q-Learning with Space Invaders

Thumbnail

AI-generated summary The text explains how Deep Q-Learning replaces a Q-table with a neural network to approximate Q-values for environments with large state spaces such as Space Invaders and other Atari environments. The curriculum covers training agents using RL-Zoo and Stable-Baselines while detailing preprocessing steps like stacking four frames and reducing state space to handle temporal and spatial properties.

Self-Supervised Vision Model Estimates Live Pig Weight Within Three

AI-generated summary Agricultural engineers and animal scientists developed an automated dual-stream computer vision framework named DINOv3-DFR that estimates the live body weight of finishing pigs from top-down overhead visual and spatial imagery. Tested on thousands of paired colour and depth image pairs collected in commercial housing, the system achieved a mean absolute error of 2.96 kilograms and a coefficient of determination of 0.927.

Also covered byScience

Gemini API Tutorial: Carbon in 12 Steps, 70 Min [2026]

Modern white and brick-red office building behind a grassy verge, with a white Google sign in front under a clear blue sky

AI-generated summary Google launched Gemini 4 Carbon for paying developers through the Gemini API with introductory pricing set through November 30, 2026. Google also released Gemini 4 Hydrogen and Gemini 4 Argon, while Stack Overflow reported that 35% of developers used Gemini Flash models.

Also covered byWeb PulseDigIntoAI
Earlier:

Australia's central bank finds AI hitting young workers first, especially in software and customer service

AI-generated summary An analysis by Mike Major for Australia's central bank indicates that artificial intelligence is affecting young workers first, particularly in software development and customer service. The study notes that job postings have declined across exposed groups since late 2022, though the researcher cautions against drawing early conclusions without direct causation.