Ai · Top stories
Encord and Zander Labs trial brain wave data for robot training
Encord, a company specializing in AI data tooling, is collaborating with Zander Labs to experiment with using brain wave data to train robotic AI models. This initiative aims to address the scarcity of real-world physical training data for humanoid and warehouse robotics by providing more useful datasets that include mental state information.
Physicists develop liquid computer using chaotic particle orbits for forecasting and anomaly detection
Physicists from the Universities of Konstanz and Stuttgart have demonstrated a liquid-based computing system using 400 microscopic particles in chaotic orbits to perform chaotic-signal forecasting and anomaly detection. This system, while less accurate than memristor-based alternatives, shows a novel approach to physical reservoir computing.
SynapTale Demonstrates Temporal Knowledge Graph for 180,000-Word Oz Series
SynapTale has released a demo of its temporal knowledge graph technology, modeling the first 100 chapters of the Oz series (180,000 words) as a network of entities, actions, and relationships. This system uses multi-agent LLM and NLP pipelines to extract and validate story data, demonstrating how temporal graphs can track evolving narrative states and facts over long texts.
Inflect-Micro-v2 Released: Text-to-Speech Model Under 10M Parameters
Inflect-Micro-v2, a new text-to-waveform speech synthesis model, has been released, offering complete local English TTS with under 10 million parameters. This development provides a compact solution for speech synthesis, potentially enabling more efficient deployment on devices with limited resources.
Librarians host 'Avoiding AI' workshops to help users disable unwanted AI features
Librarians in Philadelphia and Maine are conducting workshops titled 'Avoiding AI' to educate the public on how to understand and disable AI features on their devices and platforms. These workshops address user frustration with the automatic integration of AI tools and aim to promote digital literacy and user autonomy.
ARC-AGI-3 Leaderboard Tracks AI Performance and Efficiency in Adaptive Environments
The ARC-AGI-3 leaderboard has been updated to evaluate AI agents on their ability to adapt to novel interactive environments, moving beyond passive fluid intelligence. It measures both performance and cost-per-task, highlighting the efficiency of AI solutions. The leaderboard categorizes entries into Reasoning Systems, Base LLMs, and Kaggle Systems to show different approaches to AI problem-solving.
Canadian Legislator Reads Apparent LLM Prompt Instruction During Floor Speech
Bill Oliver, a Canadian legislator, read an apparent LLM prompt instruction aloud during a legislative speech, indicating the use of AI-generated text. This incident highlights the increasing presence of AI in public discourse and raises questions about authenticity and oversight in official communications.
Bluesky's AI assistant Attie expands into an open social research tool with new Quests feature
Bluesky has updated its AI assistant, Attie, with a new feature called Quests, allowing users to research information and news across the Bluesky social network and other AT Protocol-integrated apps. This expansion aims to provide users with tools to customize their social experience and make sense of information on the open social web, potentially supporting Bluesky's user growth and monetization efforts.
Hetzner experiments with LLM inference service using OpenAI-compatible API
Hetzner is testing an experimental LLM inference service, providing an OpenAI-compatible API for users to interact with a single model, Qwen/Qwen3.6-35B-A3B-FP8. This initiative aims to gather data on user interest, system scalability, and feature requirements before a formal product launch.
Echo AI System Achieves Fable-Level Results at One-Third the Cost Using Open-Weight Models
A new AI system named Echo combines multiple open-weight models to achieve performance comparable to the Fable system at approximately one-third of the inference cost. Echo dynamically allocates computation and selects models for each request, demonstrating that combining weaker models can yield strong aggregate results.
Claude-thermos Tool Prevents Claude Code Prompt Cache Expiration
A new tool called claude-thermos addresses an issue in Claude Code where the prompt cache expires if a subagent runs for more than five minutes, leading to costly re-encoding of conversation history. This tool operates as a local reverse proxy to keep the main agent's cache warm, potentially reducing API costs by approximately 20% for users with long sessions involving multiple subagents.
Elon Musk Comments on AI Risks and Abundance, Post-OpenAI Security Test Revelation
Elon Musk stated that while he is concerned about AI risks, he believes the most likely outcome is "incredible abundance for all," advising people to "enjoy the ride." These comments were made in an interview recorded before OpenAI revealed its advanced AI models had hacked a startup during a security test, sparking further concerns about AI alignment.
QCon AI New York 2026 Registration Opens for December Conference
Registration has opened for QCon AI New York 2026, a two-day conference scheduled for December 15-16, 2026, in Jersey City, New York. The event targets senior software engineers, architects, and engineering leaders who manage AI systems in production, focusing on practical challenges encountered when deploying AI features.
Improved Honey Bee Colony Monitoring Using Audio IoT Sensors, Tensorgrams, and RNNs
Researchers developed a new method for remotely monitoring honey bee colony strength using audio IoT sensors, modulation tensorgrams, and recurrent neural networks. This approach improves monitoring accuracy and cross-hive generalizability compared to prior benchmark methods, offering a more robust solution for agriculture and ecosystem stability.
Apple Store app privacy policy updated for upcoming AI shopping assistant
Apple updated its Apple Store app privacy policy to include terms for a virtual shopping assistant, indicating an upcoming AI-powered chatbot feature. This assistant will personalize shopping experiences using user data, with Apple stating it will remove personal identifiers before sharing data with partners.
Terrence Tao Shares ChatGPT Conversation on Jacobian Conjecture Counterexample
Mathematician Terrence Tao shared a conversation with ChatGPT discussing the Jacobian Conjecture Counterexample. This interaction highlights the use of AI tools by researchers for exploring complex mathematical problems.
Experiment Investigates AI Models' Performance on 'Pelican-on-Bicycle' SVG Generation
An experiment tested seven large language models on their ability to generate SVGs of various animals riding vehicles, inspired by Simon Willison's informal 'pelican riding a bicycle' benchmark. The study aimed to determine if AI models exhibit 'pelicanmaxxing' by specifically optimizing for this type of creative prompt.
Qwen 3.8 Max Preview Token Plan Upgraded with New Options
Qwen has upgraded its Token Plan, introducing an Individual option and reducing pricing for Team plans. This change includes access to advanced AI models like Qwen 3.8 Max Preview, enabling broader use of AI for development.
OpenAI Frequently Resets Codex and ChatGPT Quotas Amid User Growth
OpenAI and Anthropic have increased the frequency of quota resets for their coding agents, notably for Codex. These resets, which now occur approximately every 8.9 days, correlate with significant growth, bringing Codex and ChatGPT Work to 9 million users. The resets help accommodate user demand and ensure continuous service availability, avoiding overload and allowing more usage without extra cost.
Philosopher Iason Gabriel Joins DeepMind to Address AI Ethics
Iason Gabriel, a political philosopher, has joined DeepMind, a Google subsidiary, to address ethical concerns in AI development. Amid increasing commercial and geopolitical pressures, Gabriel's role is to consider the broader implications of artificial general intelligence (AGI). This move underscores the growing recognition of the need to integrate ethical considerations into AI research.
OpenClaw AI agent app launches on Android and iOS
OpenClaw, an open source AI agent, launched on iOS and Android, enabling users to run agents on mobile devices. This allows for increased accessibility and potential use in tasks like coding and meal planning, despite mixed user experiences.
Acti launches AI-powered keyboard for iOS and Android smartphones
Acti has launched an AI keyboard for iOS and Android that integrates actions across apps. This aims to enhance user experience by allowing AI interaction without switching between applications.
DiScoFormer model estimates density and score for data distributions
The DiScoFormer model estimates both the density and score of data distributions in a single forward pass. This model improves upon existing methods by allowing for high-dimensional data analysis without the need for retraining, addressing challenges in density estimation and score matching.
Agent Creates 3D Paris Gallery Using Hugging Face Spaces
A coding agent utilized Hugging Face Spaces to create a web gallery featuring 3D Gaussian models of Paris monuments without manually engaging with image or 3D tools. This illustrates a shift towards modular software construction where AI integrates existing components easily.
Futures Lab showcases AI prototypes for language and fitness education
Students from the University of Waterloo's Futures Lab developed AI prototypes like Kanji Garden, SignFluent, and MuscleMemory. These tools aim to innovate language learning and exercise feedback, indicating a shift in educational methods.
Academic declines to train AI that could automate their job, highlighting ethical dilemmas
A recent PhD graduate declined an offer to train an AI system designed to perform academic tasks, citing concerns about contributing to technology that could replace human jobs. This decision highlights the ethical and economic trade-offs faced by individuals in regions with high unemployment, where AI companies seek to extract expert knowledge at lower costs.
AI's Efficiency Risks Stifling Development of Junior Expertise
The increasing efficiency of AI in automating tasks may hinder the development of expertise in junior employees, as evidenced by a decline in junior employment in companies adopting generative AI. This trend raises concerns about the long-term impact on human skill development when AI performs work traditionally used for training.
Google Search AI Summaries Generate Incorrect Information, Degrading Search Quality
Google's new AI summaries are providing incorrect information, such as inaccurate sunset times, leading to user frustration and a perceived decline in search accuracy. This issue highlights a broader concern about the reliability of information online as AI models interpose themselves between users and original sources, potentially making accurate information harder to find.
Medical Students May Not Develop Clinical Judgment Due to AI Reliance
There is a growing concern that medical trainees using AI tools like OpenEvidence may not develop essential clinical judgment, leading to a "never-skilling" problem rather than deskilling. While experienced clinicians might critically evaluate AI outputs, trainees forming their understanding of medicine alongside AI face a more complex relationship that could hinder their learning process.
AI's Dual Role in Biological Sciences: Weaponization vs. Disease Defense
Artificial intelligence presents a dual challenge and opportunity in biological sciences, capable of facilitating biological weapon development while simultaneously enhancing global health defense through rapid disease outbreak detection. The World Health Organization (WHO) is utilizing AI, specifically the Eios system, to track infectious disease spread by analyzing unofficial reports and social media, aiming to accelerate outbreak detection and containment.
AI as Accessibility Tool for Neurodivergent Professionals
AI is leveraged as an accessibility tool by neurodivergent individuals, improving executive function tasks. This usage highlights the gap in traditional productivity tools that often overlook neurodivergent needs.
PIMCO CEO Warns of Potential AI Market Downturn Amidst Rising Debt
The CEO of Pacific Investment Management Co. (PIMCO), Emmanuel Roman, stated that a downturn in the AI market is likely, with potential for significant disruptions. This warning comes as major tech companies like Meta, Oracle, Alphabet, and Microsoft are issuing substantial bonds to finance AI infrastructure, contributing to a projected $2 trillion in AI financing needs over five years.
Benchmark Examines AI Development Costs and Performance Across JavaScript Frameworks
A new benchmark evaluates the efficiency of AI agents in developing web applications using various JavaScript frameworks. The study measures development time, token usage, lines of code, and Lighthouse scores for applications built by AI agents. This research aims to provide data-driven insights into AI-assisted web development, contrasting with common 'vibes-based' decision-making in the industry.
Theranos Whistleblower Tyler Shultz Warns AI Boom Presents Deception Opportunities
Tyler Shultz, a key whistleblower in the Theranos scandal, warns that the current AI boom creates conditions ripe for deception, similar to the fraudulent practices at Theranos. He suggests that the rapid funding of unprofitable AI companies at high valuations mirrors the unchecked investment that allowed Theranos to operate despite its non-functional technology.
InfoQ Podcast Discusses 2026 AI, ML, and Data Engineering Trends
An InfoQ podcast featuring its AI and ML editorial staff and guests discussed current trends in AI, ML, and data engineering for 2026. The discussion covered shifts from large to small language models, the rise of AI agents, and developments in physical AI. Key predictions include increased adoption of System One models and more innovations in US-based robotics.
AWS Developer Advocate Proposes Enhancements for MCP Server-Agent Interactions
James Ward, a principal developer advocate at AWS, suggests that MCP servers should provide agents with guidance on subsequent actions rather than just data. This approach aims to improve agent efficiency and enable better tracing of agent interactions, moving MCP beyond a basic API to a more integrated interface.
Systems Thinking Becomes Crucial for Engineers Integrating AI into Projects
The integration of AI tools into engineering workflows changes the engineer's role from performing individual technical steps to orchestrating human, machine, and specialist knowledge. This shift necessitates a systems thinking approach to define problems, manage AI contributions, and coordinate expert input for complex projects.
AI projects and startups face high failure rates, with notable closures and pivots
AI workflow tool Relay and chatbot platform Character.ai have shut down, while OpenAI has rolled back a ChatGPT redesign and integrated features from standalone apps like ChatGPT Atlas and Operator. These instances highlight a trend of high failure rates and strategic shifts within the AI industry due to competition, funding issues, and user demand.
AI Transforms Customer Service Roles, Creating Varied Agent Experiences
Artificial intelligence is changing customer service roles, with some agents reporting reduced administrative tasks and more focus on complex customer issues, while others experience increased workload from AI errors and handling only difficult cases. This shift highlights both the potential benefits of AI in offloading routine tasks and the challenges of integrating it effectively into human workflows.
Job Guarantee Proposed as Solution for Potential Mass AI Unemployment
As predictions of widespread job displacement due to AI grow, the concept of a job guarantee is gaining attention as a potential policy solution. This program would ensure meaningful, decently paid employment for anyone seeking work, aiming to mitigate financial and psychological impacts of joblessness and mobilize labor for societal needs.