Can AI talk you into a conspiracy? In three preregistered experiments with 2,724 Americans, people chatted with GPT-4o about a conspiracy they felt unsure about. The AI was told either to debunk it or to argue for it. * With guardrails removed, the AI was just as good at increasing
AI
Many headlines say AI chatbots "leak" your personal info. This study asks: is that true memorization, or are prompts giving the answer away? The authors find many "leaks" are driven by surface cues in the prompt. They introduce Cue-Resistant Memorization (CRM): test models under low-cue conditions
Literature reviews are slow, manual, and hard to reproduce. This paper unveils an algorithmic framework that uses Natural Language Processing, clustering, and interpretability tools to speed up screening, make choices auditable, and raise the quality of what gets included. As a test case, the authors mapped research on financial narratives—
AI
Smarter AI means knowing when to think harder LLMs can improve by thinking longer, but they rarely know how much thinking a task really needs. ROI-Reasoning trains models to plan their effort under a hard token budget, like students managing time on an exam. * Meta-Cognitive Fine-Tuning: Before answering, the model
AI
Teaching AI to Follow the Rules—While Staying Creative GANs can generate realistic images and data, but they often ignore the “rules of the world.” Enter LTN‑GAN: a new framework that blends Generative Adversarial Networks with Logic Tensor Networks, so the generator learns to produce samples that look real
AI
Decision trees (and ensembles like random forests and gradient boosting) make powerful predictions—but their reasoning can be hard to follow, which is risky in healthcare, finance, and other safety-critical settings. This paper shows how to turn those predictions into clear, formal explanations using Answer Set Programming (ASP), a logic-based
AI
TL;DR: Giving AI agents a "what happens next?" simulator doesn’t automatically make them smarter. Researchers tested whether agents built on vision–language models can use a generative world model—a tool that predicts future states—to preview outcomes before acting. * Agents rarely choose to simulate: in
AI
Building images and videos in ComfyUI is like wiring LEGO blocks — powerful, but easy to break when workflows get long and complex. ComfySearch is a new AI agent that automatically explores, reasons, and assembles ComfyUI workflows. It builds step by step, checks whether each piece runs, and adjusts on the
AI
Teaching video AIs to sweat the small stuff Text-to-video models often miss human-preferred details or waste compute learning from vague, whole-video feedback. LocalDPO, a new training recipe, targets the exact frames and regions that need improvement—no human labels or extra critic model required. * How it works: Use a real,
Robotics
Tiny robot, smart decisions—no cloud needed Researchers embedded an autonomous agent (written in AgentSpeak) into a palm-sized, two-wheeled robot that navigates a maze using only its onboard sensors and reasoning. No remote controller and no constant internet required. * Fast thinking: Each decision phase took under 1 millisecond. * Real-time results:
AI
New research examines how German criminal law could apply when generative AI is used to create child sexual abuse material (CSAM). The takeaway: legal risk isn’t limited to the user—it can reach model developers, researchers, and company representatives. The authors combine legal analysis with scenarios to show that
AI
Turning jailbreakers into the trapped Jailbreak prompts can trick chatbots into breaking rules. HoneyTrap flips the script: a deceptive, multi-agent defense that engages attackers, wastes their time, and keeps helpful answers flowing for regular users. * Threat Interceptor: spots risky prompts early. * Misdirection Controller: steers attackers into harmless honeypots. * Forensic Tracker: