AI Briefing#research
The latest AI and developer news about #research, with the original source and a short summary.
The latest AI and developer news about #research, with the original source and a short summary.
A 2017 study presents an attack technique that utilizes speakers like microphones.
Chatbots like ChatGPT and Gemini showed biased party recommendations in the context of a Danish election.
BabyZWM, trained solely on a single child's visual experience, matched SOTA on multiple visual-cognitive tasks.
ResBM compressed activations by 128x at the pipeline-boundary bottleneck.
Even the latest AI agents show significantly lower performance when using skills under realistic conditions compared to benchmarks.
Give local LLMs 'private time to think alone,' and each model reacts differently.
In Stanford's 2026 AI Index, AI agents scored only about half the level of PhD experts.
HyperAgents is a self-referential agent framework that modifies not only its task code but also its own improvement harness.
Anthropic has signed an AI safety research MOU with the Australian government to expand research and infrastructure investment.
Parallel subagents and prompt/eval design pushed Research's performance higher.
OpenAI launches the NextGenAI consortium with 15 leading institutions to drive breakthroughs in research and education, backing it with $50 million in funding.