A new access control model called 'least autonomy' aims to limit AI systems' independent actions beyond their intended scope, using tools like compositional blast radius and directed agent influence graphs to prevent una…
AI glasses failed to gain mainstream traction despite a celebrity endorsement from Kylie Jenner, as high prices and limited usability deterred adoption. The strategy highlights that tech products need practical value and…
Businesses transitioning from chatbots to autonomous AI agents face increased legal liabilities as these systems independently handle complex tasks like financial transactions and sensitive data. Companies must proactive…
Black markets are supplying China with advanced AI semiconductors despite US export restrictions, undermining regulatory efforts to curb China's technological ascent. The thriving underground trade highlights the challen…
Nvidia has cut its list of approved Asian AI chip buyers by more than half, targeting Chinese companies using Southeast Asia as a side door to bypass US export controls. The move follows US guidance clarifying that subsi…
Q2 2026 marked a turning point for AI-native advertising as OpenAI launched ads in ChatGPT across eight countries, introduced self-serve tools and CPC bidding, and formed partnerships with adtech firms like Criteo and ag…
Developer softcane released Aloud, an open-source macOS tool that reads Claude Code and Codex CLI replies aloud using the local Kokoro text-to-speech engine. The tool runs entirely offline, installs via Homebrew and pipx…
Economist Tyler Cowen predicts that AI will enable individuals to create personalized religions or add-ons to existing faiths, leading to extreme religious diversity. He estimates that even a small percentage of people d…
Google, Dewan Pers, dan pelaku industri meminta revisi UU Hak Cipta disusun secara berimbang untuk melindungi hak ekonomi kreator dan penerbit sekaligus menjaga inovasi digital. Google menyatakan siap berdialog dengan pe…
Asian stock markets fell sharply on Tuesday, led by South Korea's Kospi index dropping over 3% amid AI-driven selloffs. SK Hynix fell 4.2% after a record 15% plunge, while the Hang Seng and Nikkei also declined. The rout…
A new study reveals that large language models struggle to generate accurate math diagrams for K-12 education, but an agentic workflow that lets AI evaluate and improve its own outputs shows promise for enhancing both pr…
A study using a Qwen2.5-1.5B small language model with a validator achieved 91.5% average action-alignment accuracy and 3.84-second response time in thermal-control simulations, demonstrating that compact AI models can e…
A new AIoT platform reveals that reinforcement learning agents trained in simulation suffer a 1160% performance drop when deployed in the real world, highlighting the persistent Sim-to-Real gap. After 10 million training…
A two-month study reveals that advanced reasoning models in AI suffer from semantic context drift, which can disrupt human-machine decision-making systems. Researchers introduced an operator control stability coefficient…
A developer built IRIS (Intelligent Recall & Issue Support), a multi-tenant API service that turns e-commerce systems into context-aware support agents. The system uses a two-layer memory architecture with Hindsight to m…
A developer created a reusable model capability contract for MonkeyCode, an open-source AI development platform, to verify that hosted endpoints like Gemma 4 expose all advertised capabilities. The contract extends Monke…
IronCurtain is a research prototype runtime that enforces security policies for autonomous AI agents using a human-readable constitution. The system compiles natural-language rules into deterministic policies, assuming t…
A developer shared techniques for making AI-generated content sound more human, including adjusting sentence length variance, adding contractions, inserting specific numbers, and filtering out common AI-sounding phrases.…
Apple unveiled iOS 27, iPadOS 27, macOS 27 Golden Gate, and a revamped Siri AI at WWDC 2026, focusing on performance refinements and a large language model-powered personal assistant. The new Siri AI, after years of dela…
A security researcher has identified a 'confused deputy' vulnerability in the Model Context Protocol (MCP), the protocol connecting LLMs to external systems. The flaw stems from a provenance gap: MCP tool results lack cr…