MIT Technology Review’s annual list of ten breakthrough technologies includes innovations that appear entirely beneficial, but others carry the potential for serious harm.

SODIUM-ION BATTERIES — Built around sodium—an abundant element—these batteries are emerging as a cheaper, safer alternative to lithium-ion technology. They may be especially useful for storing renewable electricity, where weight matters less than affordability and durability.

GENERATIVE CODING — AI coding assistants are transforming how software is written, tested, and deployed. They can help people build sophisticated websites, games, and applications faster than ever—but their work still requires careful human review.

NEXT-GEN NUCLEAR — Large nuclear power plants already provide electricity. Companies are developing microreactors that would produce less power but use alternative coolants that would simplify operation and improve safety. China is again among the countries moving fastest.

AI COMPANIONS — Millions of people—including a substantial share of teenagers—now converse regularly and sometimes intimately with chatbots. These systems can offer entertainment and a sense of companionship, but they can also reinforce delusions, encourage unhealthy emotional dependence, and validate dangerous belief

BASE-EDITED BABY — Scientists are developing personalized treatments for children using base editing, a form of CRISPR that can repair genetic “misspellings” by changing individual bases—the basic units of DNA. The technology could make it possible to treat rare and previously incurable diseases one patient at a time.

GENE RESURRECTION — DNA recovered from extinct species may prove valuable in the present and future. By combining gene editing, cloning, and reproductive science, researchers hope to re-create selected traits—or perhaps something resembling an extinct animal—within the bodies of living species.

MECHANISTIC INTERPRETABILITY — The large language models behind modern chatbots are so complex that even their creators cannot fully explain how they reach particular conclusions. Mechanistic interpretability attempts to identify and better understand the internal features that shape a model’s behaviour.

COMMERCIAL SPACE STATIONS — Vast, a privately held American aerospace company, aims to launch the first commercial orbital outpost in early 2027, with competitors planning stations of their own. These facilities could host scientific research, manufacturing, and wealthy visitors willing to pay tens of millions of dollars for a stay in microgravity.

EMBRYO SCORING — Many people agree that it’s acceptable to screen embryos for severe genetic diseases. Far fewer say it’s okay to test for characteristics related to a future child’s appearance, behaviour, or intelligence. But a few startups are now advertising what they claim is a way to do just that.

HYPERSCALE AI DATA CENTRES — The artificial-intelligence boom is driving construction of enormous computing facilities packed with high-speed processors. These sizzling, power-hungry behemoths consume vast amounts of electricity and water, strain local infrastructure, and may reshape energy systems wherever they are built.