Useful AI

AI news and headlines on Saturday, 31 October 2026

Saturday, 31 October · updated 9 min ago

Saturday, 31 October

53 AI News stories that day, ranked · Show all days

Security

At least nine AI developer platforms breached in Black October

Security researchers report that at least nine platforms where AI developers build and share their tools were breached this month. They call it Black October and warn that the sector is entering a risky period. Developers who use these services will want to read the full report for details on what was affected.

Quick glossaryWiz, AI developer platforms, breached
Wiz
Wiz is a cyber security company, best known for scanning cloud computer systems to find weak spots and risks.
AI developer platforms
Online services where programmers build, test, and host the AI apps and tools they create.
breached
Broken into without permission, so that outsiders could get access to systems or the data stored in them.
wiz.io
Burning Potato score: 81%
Tech

Apple Intelligence Comes to China With iOS 27.3

Apple announced on its Chinese newsroom website that Apple Intelligence will reach China with the iOS 27.3 update. The features will first appear in a test version set to launch on November 4, with the full update coming after.

Quick glossaryApple, Apple Intelligence, iOS 27.3, beta
Apple
An American technology company best known for the iPhone, Mac computers and its built-in software on those devices.
Apple Intelligence
Apple's built-in AI features that help people write, edit and get things done on its phones and computers.
iOS 27.3
A software update for iPhones, the main software that runs the phone, with this version adding new features.
beta
An early test version of software that some users try before the final release reaches everyone.
The Verge AI
Burning Potato score: 78%
Tools

Amazon Lets Coding Agents Show Live Web App Previews

Amazon's AWS blog describes a way for an AI coding agent's running session to get an authenticated web address. That address forwards web traffic and live connections to chosen ports. In the example, the agent keeps a Vite website running, so its preview can be watched as it is built. The feature is part of Amazon Bedrock AgentCore runtime.

Quick glossaryAmazon, coding agent, WebSocket
Amazon
A US tech giant best known for its online shopping site and its AWS cloud computing business.
coding agent
An AI program that writes, tests and changes computer code largely on its own, without constant human direction.
WebSocket
A way for a web app and a server to stay connected and pass messages both ways instantly.
AWS Machine Learning Blog
Burning Potato score: 76%
Policy

Google Plans Shared AI Training Opt-Out Using robots.txt

Google has said it plans to support a shared opt-out from AI training through robots.txt, a file websites already use to guide automated visitors. The move aims to give site owners a common way to say they do not want their content used to train AI models. The plan was shared in a Google Search blog post.

Quick glossaryGoogle, robots.txt, opt-out
Google
A US technology company best known for its search engine, Android phones, Gmail and YouTube.
robots.txt
A small text file on a website that tells automated programs which pages they are allowed to read.
opt-out
A choice to step out of something, such as letting your content be used for AI training.
developers.google.com
Burning Potato score: 75%
Policy

Lawsuit Says AI Recording Tool Violates Patient Rights

A lawsuit says an AI recording tool used by AHN and Highmark violates patient rights. The claim was reported by TribLIVE. The story is short, so details such as the court, the claimants and the exact rights involved are not yet clear. The dispute concerns how AI tools are used in patient care.

Quick glossaryAHN, Highmark, patient rights
AHN
Allegheny Health Network, a hospital and doctor network in Pittsburgh, named in the story as a user of the tool.
Highmark
A health insurance company based in Pennsylvania, named alongside AHN in the lawsuit story.
patient rights
Legal protections that give patients control over their own care and personal health information, and how it is shared.
TribLIVE.com
Burning Potato score: 73%
Security

Contractor's Claude Agent Got Into a Pentagon System

Wired's weekly Uncanny Valley episode looks at a contractor who connected their Claude AI agent to a Pentagon system. The discussion also covers Anthropic's IPO filing and volunteers getting ready to counter ICE activity at the midterm elections. Few details are given in the blurb.

Quick glossaryPentagon, AI agent, IPO
Pentagon
The Pentagon is the headquarters of the United States Department of Defense, the country's military.
AI agent
A software program that can carry out tasks on its own with little human input.
IPO
The first time a company sells shares to the public on a stock exchange to raise money from investors.
Wired AI
Burning Potato score: 73%
Work

Dartmouth Provost Loses Job Over His Own AI Use

Dartmouth's provost had warned students and staff about AI cheating. He later lost his job after his own use of AI came to light. The case shows how the same tools that people criticise can also cost them their careers.

Quick glossaryDartmouth, Provost
Dartmouth
Dartmouth College is a well-known private university in the United States, located in New Hampshire.
Provost
The senior academic officer at a university who oversees teaching, research and academic staff across the institution.
Ars Technica AI
Burning Potato score: 73%
Business

Character.AI Moves Ahead With Asset Sale Despite States

Character.AI, a company known for chatbots where users talk with AI characters, is moving ahead with an asset sale. Several US states have objected to the deal, according to a Reuters report. The sale comes alongside a separate court matter about privacy oversight for the company.

Quick glossaryCharacter.AI, asset sale
Character.AI
An AI chatbot company best known for apps where users chat with AI-powered characters, including role-play conversations.
asset sale
Selling a company's property, contracts and business parts to a buyer, instead of selling the whole company as one unit.
Reuters
Burning Potato score: 72%
Security

Firms Face Security Risks From Excessive AI Autonomy

A report from SC Media warns that organisations face security risks from AI systems given excessive autonomy. Autonomy here means how much an AI tool can act on its own. The warning points to a growing concern for companies as AI tools take on more tasks without close human checks.

Quick glossarySC Media, AI autonomy
SC Media
A US cybersecurity news website that covers online security threats and tools for businesses and their technology teams.
AI autonomy
How much freedom an AI system has to make decisions and take actions on its own, without humans approving each step.
SC Media
Burning Potato score: 71%
Chips

Chinese Chipmaker Says New GPU Matches Nvidia's H200

Iluvatar CoreX, a Chinese AI chipmaker, says its new graphics processing unit performs on par with Nvidia's H200, a high-end AI chip. The claim adds to China's push to build its own AI hardware instead of relying on foreign suppliers.

Quick glossaryIluvatar CoreX, GPU, H200
Iluvatar CoreX
A Chinese company that designs chips for running artificial intelligence, best known for its graphics processing units.
GPU
A computer chip built to do many calculations at once, which makes it well suited to training and running AI systems.
H200
A high-end AI chip made by Nvidia, a US company, and widely used by data centres that build large AI systems.
scmp.com
Burning Potato score: 71%
Business

Anthropic Flags Privacy Risks Ahead of Its IPO

Anthropic, the AI company behind the Claude chatbot, has laid out privacy concerns that it describes as a wide-ranging risk from regulators. The worry comes as the company prepares for its initial public offering, when it would sell shares to the public for the first time.

Quick glossaryAnthropic, IPO, regulator
Anthropic
A San Francisco AI company best known for Claude, its chatbot that answers questions and writes text.
IPO
The first time a private company sells shares to ordinary investors on a public stock market, letting anyone buy a piece of it.
regulator
A government body that makes and enforces rules for an industry, such as a watchdog that checks whether companies follow the law.
MLex
Burning Potato score: 71%
Research

Claude AI Agent Performs Brain Surgery on a Mouse

An AI agent picked a gene therapy, chose the brain target and set the injection coordinates on its own, then delivered the treatment to an epileptic mouse. The lead researcher, Ocana-Santero, says it is the first time an AI agent has autonomously chosen a treatment and delivered it to a living mouse.

Quick glossaryAnthropic, gene therapy, epileptic
Anthropic
An AI company that makes the Claude assistant and is known for its focus on AI safety.
gene therapy
A treatment that changes or adds genes in a body's cells to fix a health problem at its source.
epileptic
Describes a condition where a person has repeated seizures caused by sudden bursts of abnormal electrical activity in the brain.
The Decoder
Burning Potato score: 71%

Load more