AI diagnoses brain tumours in minutes as scams fuel authenticity crisis

A German-led research team has developed an artificial intelligence system capable of classifying more than 100 molecular types of central nervous system tumours from routine microscope slides in a matter of minutes, potentially collapsing a diagnostic process that currently requires specialised laboratories and up to two weeks of waiting. The algorithm, named Hetairos, was trained on over 11,000 tissue sections from nearly 10,000 patients across four continents and can distinguish 102 different tumour subtypes, according to findings published in Nature Cancer. While the gold standard of molecular testing remains out of reach in much of the world, the new tool requires only standard histological images, promising to democratise precision neuro-oncology.

Viewed from London, comparable acceleration is taking hold in state-run health systems. Britain’s Department of Health has earmarked £20 million to deploy AI analysis of chest X-rays in every NHS trust by 2029, a programme that has already delivered faster lung-cancer diagnoses or all-clears to more than four million patients. A further £8 million pilot will test whether the technology can speed up treatment for heart failure and strokes across 13 NHS bodies. Health and Social Care Secretary James Murray said the tools are cutting waiting times and getting people life-saving treatment quicker. Meanwhile, Swedish researchers showed that commercial AI systems applied to mammograms can detect alterations linked to breast cancer up to six years before a clinical diagnosis, analysing nearly 89,000 examinations over a decade. Yet specialists caution that in the tightly regulated field of clinical trials, algorithms are no substitute for human judgment; they are only as good as the physicians and trial designers who deploy them.

Beyond the clinic, the same technologies are being weaponised by criminals, raising hard questions about the erosion of digital trust. In the United States, AI-generated job scams featuring fabricated voices and faces have already drained hundreds of millions of dollars from tens of thousands of victims, and security analysts in Argentina warn the fraud is metastasising across Latin America as job-hunters migrate online. At a cybersecurity conference in Córdoba, experts labelled the phenomenon a “crisis of authenticity,” noting that synthetic media and social engineering are reconfiguring the mechanics of global crime. The ease with which deepfakes can impersonate hiring managers or colleagues tests legal and social defences built for an earlier era.

That duality—between life-saving speed and the dissolution of trust—is forcing governments and societies to reckon with AI’s full implications. An Australian public service survey this year, titled “Let’s get real about AI,” is trying to cut through the hype and map genuine readiness inside the bureaucracy, echoing Stephen Hawking’s prescient warning that powerful AI could be the best or worst thing to happen to humanity. From the German lab to the Argentine fraud hotline, the technology’s trajectory will be determined less by code than by the political, ethical, and professional frameworks that shape its use. For all the brilliance of algorithms that read tumours in minutes, the hardest task remains a very human one: deciding whom to trust.