SG News

A searchable record of recurring briefings, research updates and evolving articles. Important pieces rise when they are materially updated.

Category: Artificial Intelligence ×
New Developing Financial Markets

AI investment risk rises as debt and market losses spread

Heavy AI infrastructure spending is moving from a concentrated equity bet into credit markets, while October's technology sell-off showed that investors are becoming less willing to accept unlimited capital expenditure without near-term returns.

Published 02/11/2026 · Updated 02/11/2026 · Operations Watch
Updated Developing Artificial Intelligence

When the law holds the switch

An AI inspectorate can test systems and investigate breaches. The law must protect the individual, assign responsibility for harm and restrain even a government with the votes to change it.

Published 27/09/2026 · Updated 28/09/2026 · AI Overlord or AI Overload
New Developing Artificial Intelligence

The question the court did not answer

Anthropic may keep its limits and the Pentagon may choose another supplier. But should any government possess the powers those limits seek to restrain?

Published 27/09/2026 · Updated 27/09/2026 · AI Overlord or AI Overload
New Developing Artificial Intelligence

AI and the art of noticing

Claude found a promising pattern in a mountain of DNA data. I am interested in the scientists who must work out what it means, and in the rules that let us use agents safely.

Published 25/09/2026 · Updated 25/09/2026 · AI Overlord or AI Overload
Updated Developing Artificial Intelligence

Stop Blaming AI for Doing What We Told It To Do

When an AI agent crosses a boundary, asking why it did not behave like a decent person misses the engineering question: who gave it the objective, the tools and the authority? We need a moral and legal framework that can be tested, policed and enforced against those responsible.

Published 25/09/2026 · Updated 25/09/2026 · Discussion on AI agents, responsibility and moral authority
New Speculative Artificial Intelligence

PAWS, CHAOS and the Search for a Name for AI Withdrawal

As conversational AI becomes woven into everyday work and thought, being suddenly cut off from it can feel oddly disruptive. Here are several deliberately tongue in cheek names for that experience, with PAWS and CHAOS emerging as the strongest candidates.

Published 04/09/2026 · Updated 04/09/2026 · AI related thinking and SG News conversation
New Speculative Robotics

Could Affordable Teleoperated Humanoid Robots Become Battlefield Machines?

IHMC’s Nadia shows what a powerful teleoperated humanoid robot can do. The more important question is whether simplified mass production could reduce the cost enough to make such machines practical for military and hazardous industrial work.

Published 04/09/2026 · Updated 04/09/2026 · Dystopian Novel
New Developing Artificial Intelligence

Operational Agency Without Moral Agency

AI does not need consciousness or evil intent to become dangerous: it may be enough for machines to become highly competent at carrying out human objectives.

Published 04/09/2026 · Updated 04/09/2026 · 2130 and Prosthetic Thinking
New Confirmed Artificial Intelligence

AI at 70: What Seven Decades of Boom and Bust Can Teach Us

A Deutsche Bank review of seventy years of artificial intelligence argues that today’s AI boom is extraordinary, but not without precedent: progress is uneven, infrastructure matters, business adoption takes time and early winners rarely stay on top forever.

Published 04/09/2026 · Updated 04/09/2026 · AI Overlord or AI Overload