A high-severity flaw in Amazon Q Developer let a malicious repository run commands and steal a developer’s cloud credentials. The path was short: a developer opens the repo, trusts the workspace, and Amazon Q does the rest. Amazon has patched it.
Tracked as CVE-2026–12957 (CVSS 8.5), the bug sat in how Amazon’s AI coding assistant handled Model Context Protocol (MCP) servers.
Wiz Research, which found and reported it, showed that a single config file dropped in a repo was enough to go from git clone to cloud compromise.
The generative AI boom is fueled by staggering investments (including OpenAI’s multibillion-dollar chip deals), but for many companies, profitability as a result of these investments has remained elusive, leading some economists to warn of an AI bubble. In this Q&A, Harvard Business School’s Andy Wu wades through the potential and hype of the new technology. In particular, he highlights structural challenges facing most companies and warns of inevitable expiration dates on current legacy subscription models. He says that the industry’s future will depend on sustainable economics and business models that are able to capture value.
When a standard large language model (LLM) is confronted with a problem, it tries to solve it by matching it to similar information it has seen before, and then give an answer based on those past patterns. But how it decides which information to use and what value it gives to different pieces of information can be somewhat inscrutable from the outside. An EPFL team has created a new large language model that is structured similarly to a human brain, allowing users more control and moving away from “black box” AI.
The LLM MiCRo (Mixture of Cognitive Reasoners) is architecturally divided into four specialized areas that act like different parts of the human brain, allowing users to have more control over how it approaches a question and to better understand how it comes to its answers. The model, which was presented at the International Conference on Learning Representations (ICLR 2026), comes from the NLP Lab, part of the School of Computer and Communication Sciences (IC), and the NeuroAI Lab, part of IC and the School of Life Sciences at EPFL. The paper is posted to the arXiv preprint server.
Working in connectomics means creating comprehensive maps of brain and nervous system networks. Your research includes the identification and measurement of all parts of each neuron: the soma, dendrites, axonal path and branching patterns and combining that data with the synapses and gap junctions of the entire circuit.
Your microscopy challenges are extensive; submicron resolution is required over long distances inside large volumes of dense and complicated tissues.
Known by acronyms that need no explanation, viruses like COVID, SARS and Ebola conjure images of medics in protective suits and spark fear in populations worldwide.
Vaccines for individual viruses have provided some relief, but new strains pose a constant challenge.
Now, new AI-aided vaccine technology developed by scientists at Cambridge University offers potential immunity against whole families of viruses and could even prevent the next pandemic, according to researchers.
Every echocardiogram is a moving story. For a baby born with a complex heart condition, the gray and black images on the ultrasound screen can influence some of the earliest and most important decisions a medical team makes: What exactly is wrong with the heart? How urgent is surgery? What should doctors watch for after repair?
In our recent work, we focused on tetralogy of Fallot, often shortened to TOF. It is one of the most common cyanotic congenital heart defects. The condition involves several structural abnormalities of the heart, and many children with TOF need careful evaluation, surgery and long-term follow-up. The research is published in the journal eBioMedicine.
Echocardiography is central to that process. It is widely used, noninvasive and rich in clinical information. But it is also demanding. Clinicians must identify the correct views, interpret moving images, measure small cardiac structures, and combine these pieces of information with the patient’s clinical course. Even experienced clinicians can face heavy workloads, and interpretation can vary between observers.
Fourteen years ago, I sat down in Ray Kurzweil’s office in Boston, fumbled with a slipping lavalier mic, and asked the man whose book pulled me into this whole world a deceptively simple question: Can we reverse-engineer the human mind?
What strikes me now, rewatching this, is how little the core debate has aged. Back in 2012, we argued about Watson, the Turing Test, whether AI deserves rights, and whether a machine would ever care about humanity’s hardest problems. Swap a few names, and that is the front page today.
But the line that has stayed with me all these years was not about #technology at all. When I asked Ray how a kid decides at age 5 to become an inventor, his answer ran counter to every productivity guru on the internet:
“Do not be too concerned about what is practical. Follow your passion and be who you would like to be.”
Coming from one of the most relentlessly practical inventors alive, the man behind the flat-bed scanner, text-to-speech, and the music synthesizer, that is not soft advice. It is a thesis about #innovation itself.
There is a reason I keep coming back to this conversation when people ask me about the #singularity and #ArtificialIntelligence. Ray’s optimism is famous. What gets missed is where he aims it.
