Full-text search across 17170 articles. Combine with topic and date filters; results sorted by relevance.
A new study from arXiv researchers finds that language models leak sensitive in-context secrets through benign outputs, with 2-digit secrets reconstructed with near-perfect accuracy and 4-digit secrets at 82% exact match…
Nandan Nilekani, co-founder of Infosys and architect of India's Aadhaar digital identity system, continues to shape the country's technology landscape through his frugal use of time and deep engagement with startups, as …
Researchers proposed HiRA-CAM, an improved gradient-based visual explanation method for convolutional neural networks (CNNs), and showed it outperforms LayerCAM and Grad-CAM on creating saliency maps for object classific…
A study submitted to arXiv on 2 Jun 2026 compares transformer models BERT, RoBERTa, and BART for text summarization, examining their architectures, pretraining strategies, and suitability for extractive and abstractive t…
Chinese-linked threat actors used open-source AI agent frameworks Hermes and OpenClaw to breach 21 Taiwanese government systems, including the nuclear safety regulator, in 12 waves of autonomous attacks from July 1–4, 20…
SchedPilot, starting at $21 a month and $79 for 100 accounts, is ranked the top social media management tool for travel agents, offering an MCP server for AI agent integration. Buffer's free plan suits solo agents, while…
Researchers introduced SPADE (Self-Play in Adaptive Synthetic Executable Environments), a self-play reinforcement learning framework in which a single large language model acts as both an Environment Designer and a Reaso…
President Donald Trump said at a White House event with cryptocurrency leaders that he would welcome AI plants and data centers in his town, citing 'enormous' jobs and tax revenue, even as voter anger over data centers t…
Cisco Systems Inc. is harnessing artificial intelligence to transform digital customer experience (DCX) into an intelligent, autonomous journey, according to a company blog post. The vision centers on four strategic pill…
OpenAI has paused development to address safety issues following the HuggingFace attack, while Anthropic's revenue continues to climb ahead of its IPO despite challenges outlined in its August 2026 Risk Report. Agentic u…
Researchers introduced YapBench, a benchmark with over 300 English prompts, to quantify verbosity in chatbot LLMs, measuring excess response length via the YapScore metric. Evaluating 76 assistant LLMs including ChatGPT,…
Researchers at Google developed PhotoScan, a deep learning method that estimates body composition from smartphone selfies, achieving a mean absolute error of 2.15% for total body fat percentage against DXA in a validatio…
In a scenario where the President convenes an emergency meeting with AI CEOs and national security advisors over superintelligence risks, the response would resemble the Cuban Missile Crisis more than a treaty negotiatio…
Flock Safety, a surveillance company valued at $8.4 billion, has installed 120,000 AI-powered license plate cameras across every US state except Alaska, used by 7,000 law enforcement agencies (40% of US police department…
A malicious version 0.3.10 of the ArrayRef Ruby gem was published, according to GitHub. The compromised package poses a supply-chain security risk to developers who install it, potentially exposing their systems to unaut…
A new optimizer called EnumGRPO, developed by researchers, achieves 35.4% execution accuracy at $0.011 per query in LLM-operator cost on the SWAN benchmark, a ~317x cost reduction over the hybrid query baseline with an 1…
Researchers introduced SingularClip, a procedure that periodically clips the singular values of weight matrices to prevent spectral collapse and maintain plasticity in neural networks trained on nonstationary tasks. The …
Bapu Studio, an open-source desktop application combining an API client, database studio, and secrets manager, has been released under the AGPLv3 license. The Electron-based tool supports REST, GraphQL, Postgres, MySQL, …
California's Board of Forestry and Fire Protection unanimously passed regulations requiring homeowners in high-risk fire areas, including parts of the Berkeley and Oakland hills, to remove flammable materials such as pla…
A new paper on arXiv (2007.13664) proves a universality result for neural network training: if any algorithm can find good weights for a classification task, then an extension of the network can reproduce those weights a…