A new study explores agentic AI for detecting and characterizing immune-related adverse events from clinical notes, highlighting its potential to support faster and more consistent immunotherapy toxicity surveillance.
Category: robotics/AI – Page 24
AIs fail to reliably detect AIs in reverse Turing tests
The famous “Turing test” of computer science comes from a 1950 paper written by the British computer scientist (and war hero) Alan Turing. There are many variants, but perhaps the most common involves a human asking questions of “someone” behind a screen. The test is: Can the human consistently determine if the answerer is a human or a machine?
A new paper reverses this situation and asks: Can a large language model (LLM) ask questions that reveal if the hidden respondent is a human or another LLM? The work, by four French scientists from the Unit of Artificial Intelligence of PITOR at Paris Cité University, is published in the journal Computers in Human Behavior Reports.
Enabling Multi-Agent Penetration Testing Environments for Software Vulnerability Detection
In light of the ongoing challenge posed by unknown vulnerabilities in software solutions, the prompt identification and rectification of these weaknesses are of vital importance. Traditional penetration testing, although thorough, typically demands substantial human resources and time commitments. To contend with this issue, we propose an innovative penetration testing framework that incorporates natural language processing, utilizing large language model (LLM) driven multi-agent systems to automate and expedite vulnerability detection. By integrating insights from established exploit signatures and AI-driven fuzzing techniques, our framework, LLMAZD, preemptively examines target software within a controlled setting, effectively identifying both zero-day and one-day exploits, as well as system vulnerabilities, in a cost-effective manner. The framework demonstrates versatility in detecting multiple vulnerability classes, including SQL injection, Cross-Site Scripting (XSS), Cross-Site Request Forgery (CSRF), file upload vulnerabilities, and other different types, all while maintaining efficiency and cost-effectiveness throughout the testing process. Extensive evaluations on more than 20 CVE test cases reveal that the framework substantially enhances detection capabilities, reduces the overall burden of testing, and offers a practical pathway towards more secure software ecosystems.
Cyber-Security, Large Language Models (LLMs), Zero-Day Exploit, Malware, Vulnerability, Exposure, Multi-Agent Systems, Penetration Testing
Juan Enriquez on Right/Wrong: How Technology Transforms Our Ethics
In October 2020, two years before ChatGPT, I sat down with Juan Enriquez for 90 minutes to talk about right and wrong.
Not tech #ethics. Ethics.
Juan’s argument is unsettling in the best possible way: much of what we feel certain is right today rests on the technology we happen to have. Change the technology and the moral ground shifts underneath us, usually before anyone notices.
Which raises a question I still can’t shake. If our grandchildren will judge some of our certainties the way we judge our grandparents’, which ones will it be? And what happens when we hand moral decisions to machines trained on today’s certainties?
Juan has given nine TED talks, sailed around the world on an expedition that multiplied the number of known genes a hundredfold, and sat on the peace commission that negotiated a ceasefire with the Zapatistas. He also calls himself a very curious and optimistic Cro-Magnon.
We went from Catholic ethics to humility and forgiveness, from homo evolutis and #transhumanism to veganism, techno-solutionism, #AI and the singularity. Plus why those who can make you believe absurdities can make you commit atrocities.
Country of Geniuses in a Data Center 📄
HELIOX: WHERE EVIDENCE MEETS EMPATHY 🇨🇦 # *Country of Geniuses in a Data Center **📄*
https://youtu.be/XLGDB5C5qms](https://youtu.be/XLGDB5C5qms)](https://youtu.be/XLGDB5C5qms)
*Inside Anthropic’s S-1: rogue agents, an $80 billion paper loss, and who really holds the brake*
*Nov 1, 2026 • S8 E11 • 48:48*
An SEC filing is supposed to be dull. Anthropic’s S-1 reads like a confession, a safety manual, and a geopolitical thriller at once. 📄
New Heliox episode: Country of Geniuses in a Data Center. Inside Anthropic’s S-1: rogue agents, an $80 billion paper loss, and who really holds the brake.
In this deep dive, we decode the October 2026 IPO filing from the maker of Claude:
New neural framework could make LLMs more empathetic
As more people turn to large language models (LLMs) during personal crises, the companies behind them are engineering these AI-powered models to be more personal, empathetic and understanding when prompts indicate a user is in emotional distress.
The trend is raising concern about how well prepared ChatGPT, Claude, Gemini and other popular LLMs are to provide appropriate emotional support.
To address this concern, Georgia Tech researchers have developed a new neural framework that examines how different training sources influence empathy and social reasoning within these AI-powered models.
One-shot design of functional protein binders with BindCraft
I read the BindCraft paper: super cool approach for using AI to design de novo protein binders with a high success rate! While RFD4 Proteina seems to have recently outdone BindCraft in a lot of ways, this was still a very interesting and educational read.
Computational protein design offers a powerful alternative, enabling the tailoring of binders to specific targets and binding sites. Physics-based methods such as Rosetta allowed early binder design by means of scaffolding and side-chain optimization2,3. However, such methods suffer from low experimental success rates (less than 0.1%) and require the sampling of many designs2,4. Moreover, they typically require the docking of predefined scaffolds onto a fixed target structure, leading to surface incompatibilities and suboptimal binding, or even precluding the targeting of certain sites.
Recent breakthroughs in deep learning have revolutionized the field of biomolecular modelling. Models such as AlphaFold2 (AF2)1 have demonstrated remarkable capabilities in accurately predicting protein structures and complex PPIs. Indeed, AF2 filtering has been shown to increase the success rates of binder design by evaluating the plausibility of predicted complexes4,5. Deep learning has also been successfully applied for de novo design of proteins and binders. The current state-of-the-art methods involve the use of RFdiffusion5 for backbone generation coupled with ProteinMPNN sequence generation6. When applied to binder design, this approach shows significantly improved success rates compared with previous methods5. However, RFdiffusion relies on sequence design over side-chain-free backbones placed at a rigid target interface, with binder selection ultimately depending on AF2-based complex prediction to identify plausible interactions. This highlights a gap between backbone generation and functional interface design that AF2 filtering helps to bridge.
Given the use of AF2 in improving binder filtering success, we proposed that we could harness it directly for the design of protein binders. We present BindCraft, a user-friendly pipeline for de novo design of protein binders that requires minimal user intervention and computational expertise. BindCraft leverages backpropagation through the AF2 network to efficiently hallucinate new binders and interfaces (Fig. 1a). We demonstrate the efficiency of our pipeline on 12 diverse, challenging and therapeutically relevant protein targets (Fig. 1b). We identify several high-affinity binders for each target without the need for high-throughput screening of hundreds to thousands of designs experimentally. This marks an important advancement in the design of protein binders on demand, and makes binder design accessible to research groups without expertise in computational design methods or access to high-throughput screening facilities.
Indian AI infrastructure firm reportedly seeks to raise $3 billion of debt to buy Nvidia’s next-generation chips
Nvidia’s Rubin GPUs are tailored for agentic AI and will give AM Intelligence a first-mover advantage in India, Shah said.
By next year, enterprises will be focusing on “running low-latency production agents” and not just on the raw foundational model training, which is supported by Blackwell Ultra GPUs.
AM Intelligence is developing 5 GW of AI data centers and 1 GW of compute-as-a-service across India, the U.S. and Europe, a spokesperson for the company said in a note. Within two years, the company plans to set up 400 MW of compute capacity.
The Emerging Technologies of Our New Era: How Innovation Will Transform the Human Ecosystem
Humanity is about to enter a new technological era that has the potential to drastically alter civilization, much like the Industrial Revolution, the discovery of electricity, and the development of the internet. However, unlike earlier times of change, the advancements of today are not happening on their own. The convergence of artificial intelligence, quantum computing, biotechnology, advanced materials, robotics, space systems, and next-generation communications is producing capabilities that were remarkably difficult to envision just a generation ago.
The outcome is more than just a set of novel technologies. It is a developing ecosystem where humans, biological systems, intelligent machines, and interconnected infrastructure will interact in ever-more complex ways. Technological advancement brings both remarkable opportunities and consequential risks throughout my writing on cybersecurity and emerging technologies, including my book “Inside Cyber: How AI, 5G, IoT, and Quantum Computing Will Transform Privacy and Our Security.”
Microexons May Help Explain the 85% of Autism Cases Without a Clear Genetic Cause
Researchers used machine learning and large-scale genomic data to identify new neuronal microexons and rare variants affecting their splicing. The findings may explain some previously unresolved genetic contributors to autism.








