A developer behind the MCP security scanner WARDEN benchmarked it against two open-source scanners, mcp-audit and mcp-shield, on 485 poisoned tool definitions extracted from the MCPTox dataset and on …
A teacher-guided low-fidelity fitness proxy combined with constrained evolutionary hardening and a trust-region sequential quadratic programming backbone can cut evaluation cost by more than 50% while…
A technical post in the Machine Learning series demonstrates teaching the sw-MLPL array language to express convolutional neural network operations, using the triple-sum convolution definition from Zh…
Prefix caching can push LLM prefill from compute-bound to memory-bandwidth-bound once enough of the prefill is loading cached KV rather than computing it, according to a preprint by Meng, Lee and Wang…
A developer argues that AI coding agents should be auditable through execution evidence rather than trusted based on their reports. The post distinguishes between requests and results, proposes a thre…
Mathematician Terence Tao released a web app that visualizes the logical flow of mathematical papers as interactive diagrams, using a JSON format to represent dependencies between theorem objects. The…
A new analysis from ai-2040.com explores five 'Plan A' scenarios for US-China cooperation to slow transformative AI development, including chip-level compute control, a joint international project mod…
Chinese college application consultants are using fake credentials, AI-generated advice, and predatory marketing to lure high school graduates and their families into spending thousands of yuan, accor…
A developer created a 40-line Python script called context_tax.py to measure the 'context tax' in AI agent sessions, where each step re-sends the entire conversation history as input, causing total to…
A PhD student released CMG, an open-source tool that audits LLM-based judges by requiring them to back each verdict with explicit claims tied to evidence, flagging untrustworthy decisions without usin…
Taiwanese prosecutors charged three individuals with forging export documents to reroute approximately 50 Supermicro AI servers, valued at over $15 million, to mainland China via Japan. The suspects a…