Full-text search across 7151 articles. Combine with topic and date filters; results sorted by relevance.
OpenAI's Codex tool now encrypts MultiAgentV2 message payloads end to end, preventing local audit trails from showing what parent agents asked child agents to do. The change, merged into main on 2026-06-05 and included i…
A new guide outlines 12 practical strategies for reducing latency and inference costs of large language models (LLMs) in production, emphasizing that most gains come from eliminating unnecessary work rather than adding m…
Autonomous agents using large language models (LLMs) struggle to replicate recent scientific findings, with top systems achieving less than 50 F1 on the new FIRE-Bench evaluation framework, which tests full-cycle scienti…
A new mechanistic interpretability study from researchers finds that LLM-as-judge scoring bias has a geometric representation in the model's hidden state, enabling causal control and operational prediction. Across seven …
Researchers have developed modelDNA, a tool that verifies the lineage of open-weight large language models by fingerprinting a model from roughly 100-300 MB of ranged HTTP reads and comparing it against a reference datab…
AI coding agents should optimize for less owned code to avoid technical debt, argues a new analysis, as the cost of code generation drops but ownership costs rise. Black Duck's 2026 report found that 97% of audited comme…
Researchers seeking endorsement for arXiv submission of their manuscript "Nova: An End-to-End MLIR Compiler for Deep Learning" request an established researcher to review and endorse their work in categories cs.LG, cs.AI…
Augmodo, a Seattle startup that sells a dual-camera wearable badge for retail workers, raised $21 million in a round led by TQ Ventures at a $350 million valuation, with plans to expand into warehouses, factories and hos…
The FDA's January 2025 draft guidance on AI-enabled medical devices recommends that model descriptions detail inputs, outputs, architecture, training, and limitations, but the document remains draft as of July 2026. In p…
VRAM capacity and memory bandwidth, not AI TOPS, are the decisive specs for local LLM inference in mid-2026, according to an analysis by Ji-ho Choi. NVIDIA's RTX 5090 leads with 32 GB VRAM and 1792 GB/s bandwidth, while …
AuditWeave, a lightweight Python library with no runtime dependencies, records AI-assisted and data-transformation workflows into an append-only, hash-chained ledger that enables tamper detection and end-to-end traceabil…
A study comparing Turbo-Quant and SpectralQuant KV-cache compression found that eigenbasis-based methods fail on heavy-tailed data due to covariance instability but excel in structured regimes, with the effective semanti…
A new book by computer scientist Peter J. Denning argues that artificial intelligence research has been built on a flawed assumption from Alan Turing's 1950 paper, making true human-level AI impossible. Denning claims th…
The FBI reported roughly 59,000 romance scam victims in 2024, with generative AI making scams more convincing through deepfake videos and voice clones, according to The Women's Tech & Telecom Partnership's 2026 Impact Re…
Andrej Karpathy's LLM Wiki method uses Obsidian and Claude Code to build an AI-maintained knowledge base, turning passive notes into a living document that a language model can read, update, and reason over. The approach…
Skyfall AI released MORPHEUS, a persistent enterprise simulation benchmark for continual reinforcement learning that requires agents to learn under structured non-stationarity without environment resets. The platform use…
USC assistant professor Ibrahim Sabek has received a $627,250 NSF CAREER Award for a five-year project titled "Toward Quantum-Augmented Database Systems" that aims to integrate quantum computing into database engines to …
OpenAI released GPT-5.6-Sol, alongside cheaper models Terra and Luna, pricing Sol at $5/$30, Terra at $2.50/$15, and Luna at $1/$6. CEO Sam Altman said the models address enterprise cost concerns and deliver a huge step …
Apple released the watchOS 27 public beta for Apple Watch, featuring Siri AI, a dynamic app grid, a one-handed tap gesture, and a more proactive Smart Stack. The update requires an Apple Watch Series 9 or newer and an iP…
An Android developer explains how to implement always-on audio AI using a Digital Signal Processor (DSP) to avoid battery drain and overheating. The post details the hardware paradox where CPUs are too power-hungry, GPUs…