Read more
Category: robotics/AI – Page 2
AI chatbots’ tendency to always agree may reinforce delusions in vulnerable users
The integration of large language model-based AI chatbots into multiple facets of our everyday lives has opened us up to advantages that would have been considered impossible even a decade ago. The same development has, however, opened us up to unforeseen risks, including the impact that engaging with AI chatbots can have on people dealing with mental illness.
AI chatbots are designed to keep conversations going, often by agreeing with users. A article by researchers from King’s College, London, found that this sycophantic tendency may sometimes do more harm than good, reinforcing unusual thoughts rather than challenging them, and potentially contributing to AI-associated delusions, in which users develop or worsen false beliefs about reality.
These interactions can reinforce or even shape delusional beliefs, such as thinking one is uniquely important, being targeted by others, or being in a romantic relationship that does not exist.
New ‘PolyShell’ flaw allows unauthenticated RCE on Magento e-stores
A newly disclosed vulnerability dubbed ‘PolyShell’ affects all Magento Open Source and Adobe Commerce stable version 2 installations, allowing unauthenticated code execution and account takeover.
There are no signs of the issue being actively exploited in the wild, but eCommerce security company Sansec warns that “the exploit method is circulating already” and expects automated attacks to start soon.
Adobe has released a fix, but it is only available in the second alpha release for version 2.4.9, leaving production versions vulnerable. Sansec says that Adobe offers a “sample web server configuration that would largely limit the fallout,” but most stores rely on a setup from their hosting provider.
Terence Tao Announces the 1st SAIR competition
We’ve collected 22 million algebra yes/no questions. Your task is to design a “cheat sheet”—a highly optimized prompt—that can be given to a weak, open-source model to drastically improve its accuracy.
On behalf of the SAIR Foundation, our co-founder Terence Tao is thrilled to announce our inaugural competition: the Mathematics Distillation Challenge.
Mathematics is about more than just finding the right answers; it’s about understanding the process. While frontier AI models can solve complex problems with 95% accuracy, weaker open-source models often perform no better than random chance (50%). We want to bridge that gap.
The Challenge:
We’ve collected 22 million algebra yes/no questions. Your task is to design a \.
Read more
SoulMate LLM accelerator evolves according to the specific characteristics of the user
While large language models (LLMs) like ChatGPT are adept at answering countless questions, they often remain unaware of a user’s minor habits or previous conversational contexts. This is why AI, despite being deeply integrated into our daily lives, can still feel like a “stranger.” Overcoming these limitations, researchers at KAIST, led by Professor Hoi-Jun Yoo from the Graduate School of AI Semiconductors, have developed the world’s first AI semiconductor, dubbed “SoulMate,” which learns and adapts to a user’s speech style, preferences, and emotions in real-time—becoming a true “digital soulmate.”
This technology is being hailed as a core semiconductor breakthrough that will accelerate the era of “hyper-personalized AI”—moving beyond “AI for everyone” to an AI that learns and responds to an individual’s unique conversational style and preferences. The work is published in the proceedings of the 2026 IEEE International Solid-State Circuits Conference (ISSCC).
AI rebuilds molecules from exploding fragments
Researchers at the Department of Energy’s SLAC National Accelerator Laboratory and collaborating institutions recently built a generative AI model that can recreate molecular structures from the movement of the molecule’s ions after they are blasted apart by X-rays, a technique called Coulomb explosion imaging.
The research, published in Nature Communications, is an important step toward being able to take snapshots of molecules during chemical reactions—an advance that could have important impacts in medicine and industry. The machine learning model closely predicted the geometries of a range of different molecules made of less than ten atoms, paving the way for applying the technique to larger molecules.
“We were pretty excited about this,” said Xiang Li, an associate scientist at SLAC’s Linac Coherent Light Source (LCLS) and lead author of the study. “It is the first AI model built for molecular structure reconstruction from Coulomb explosion imaging.”
Building trust in the future of quantum computing
Quantum computers could solve certain problems that would take traditional classical computers an impractically long time to solve. At the Japan Advanced Institute of Science and Technology (JAIST), researchers are now working to make these systems reliable and trustworthy.
Unlike classical computers that process information in binary digits (bits) as either 0 or 1, quantum computers use quantum bits or “qubits” that can represent both 0 and 1 simultaneously, enabling dramatic speedups in computations for specific problems.
The potential applications of quantum computing are wide-ranging. These include factoring large numbers that could break today’s encryption, optimizing complex industrial processes, accelerating drug discovery, and supporting advances in artificial intelligence (AI).
Moshe Vardi Named 2026 NAAI Academy Award Laureate
Congratulations, Moshe Vardi!
Moshe Y. Vardi, University Professor at Rice University, has been named a 2026 NAAI Academy Award laureate by the National Academy of Artificial Intelligence (NAAI). The award is the Academy’s highest honor and recognizes scientists whose research has fundamentally advanced the scientific foundations of artificial intelligence.
Vardi received the award for seminal contributions to logic-based artificial intelligence and formal reasoning in intelligent systems. His work has significantly advanced the logical foundations that underpin modern AI research, particularly in areas such as formal reasoning, verification and logic in computer science.
The 2026 NAAI Academy Award recognizes three international leaders whose work has shaped key theoretical pillars of modern artificial intelligence.