Can technology really become dangerous enough to destroy humanity?

We have already crossed that bridge once.

When scientists learned how to release the energy contained within the atomic nucleus, humanity acquired a technological capability powerful enough to threaten civilization itself. Nuclear weapons did not end technological progress, but they permanently changed the relationship between innovation and consequence. Once humanity possessed them, we could no longer pretend that every technological failure would be local, recoverable, or even survivable.

What Dennis Prager’s philosophy of moral authority can teach us about AI alignment

By Dennis C. Hayes

An artificial-intelligence model does not pray, worship, fear divine judgment, or necessarily experience conscience. It may be able to discuss God with extraordinary sophistication, but that does not mean it believes in God—or believes anything in the human sense.

Why, then, should a book about God and morality have anything to teach us about artificial-intelligence alignment?

What Evolving Microprocessor Programs Reveal About Cooperation, Alignment, and the Architecture of Human Survival

A population of computer programs begins with no intelligence, no social rules, no designed method of reproduction, and no instruction to cooperate. Each program must spend scarce energy to execute its code. It can preserve the shared energy on which both programs depend, or it can steal energy from another program for an immediate advantage. Given enough time, which behavior survives?

Abstract

Artificial intelligence security faces a fundamental long-term problem. Today, the most capable AI models require enormous computational resources, specialized hardware, sophisticated engineering teams, and large capital investments. These requirements provide a temporary form of concentration: only a relatively small number of organizations can develop and operate frontier systems. That concentration makes governance, monitoring, evaluation, and intervention possible.

It should not be assumed to last.

Something important is changing in the American debate over artificial intelligence and the enormous physical infrastructure required to support it.

On September 8, 2026, OpenAI announced something that, if confirmed by the mathematics community, may become a landmark not only in mathematics but in the history of artificial intelligence: an AI-driven research effort produced what OpenAI says is a solution to the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems.

The result is important. But the way the result was produced may be equally important.

The Good News About America’s Data-Center Backlash: The Myths Are Finally Being Challenged

The good news is that even a newspaper generally identified with the liberal side of American political debate, The Washington Post, is beginning to challenge some of the myths, exaggerated claims and bad arithmetic driving opposition to AI data centers.

A research team affiliated with Google Research, Cornell University, and the University of Southern California has introduced a promising new method for improving how artificial intelligence models remember long sequences of information. The paper, “Memory Caching: RNNs with Growing Memory,” was written by Ali Behrouz, Zeman Li, Yuan Deng, Peilin Zhong, Meisam Razaviyayn, and Vahab Mirrokni. It was accepted for presentation at the 2026 International Conference on Machine Learning.

The General Architecture of Technological Consequence
A Common Framework for Understanding Innovation, Risk, and the Governance Gap

A Technical Review of a Provocative New Model of Human–AI Interaction

A provocative new research paper asks whether large-language models should be understood not merely as tools, but as entities that spread through society, attach themselves to human cognitive processes, and potentially create persistent dependence.