The news desk
Recent developments in artificial intelligence, with the original sources one click away.
Researchers introduce ConfidenceBench to evaluate confidence calibration in LLMs, while separate studies assess watermarking risks in medical AI and lightweight model performance.
Sources · arxiv.org
Jefferies implemented an AI-powered trade assistant using AWS Strands Agents SDK to enhance front office trading operations.
Sources · Amazon AWS AI
Meta published an AI optimism ad set to Bowie's 'Five Years,' a song about human extinction, drawing attention to the ironic soundtrack choice.
Sources · wired.com / techcrunch.com
New research shows targeted training approaches can enhance small language models' performance on domain-specific languages and reasoning tasks.
Sources · arxiv.org
Four new research papers evaluate LLMs on scientific visualization, security testing, network automation, and combinatorial coding tasks.
Sources · arxiv.org
OpenAI unveils Project Camellia in Georgia with commitments to responsible energy, community investment, and local job creation.
Sources · OpenAI
New studies examine how reasoning capabilities in language models affect translation quality and computational costs.
Sources · arxiv.org
ServiceNow makes strategic investment in BusinessNext to expand AI-powered banking software globally and strengthen financial services presence.
Sources · TechCrunch AI
The Trump administration is internally debating policy approaches to address the growing capabilities of Chinese AI models.
Sources · WIRED AI
Material science innovations underpin AI progress in algorithms, computing, and infrastructure, according to industry analysis.
Sources · MIT Technology Review
Google launched Gemini 3.6 Flash, 3.5 Flash-Lite, and Flash Cyber models, competing with OpenAI and Anthropic in the AI race.
Sources · NYT Technology / TechCrunch AI
Recent arXiv papers examine how reasoning capabilities affect translation quality and computational costs in neural machine translation systems.
Sources · arxiv.org
OpenAI introduces new program to help small business owners build AI skills and automate work using ChatGPT Work.
Sources · OpenAI
Research reveals that large language models can develop biases beyond those in training data when screening job candidates.
Sources · MIT Technology Review
New research platform enables LLM agents to construct editable data pipelines with 93.3% success rate and 72.5% cost reduction.
Sources · arxiv.org
Google's integration of AI into search results is increasing user time on Google's site, prompting complaints from website operators.
Sources · NYT Technology
OpenAI details new safety risks and improved safeguards discovered through deploying long-running AI models in real-world settings.
Sources · OpenAI
New research introduces TRACE, a safety patch learning method that maintains model utility while preventing harmful outputs in customized LLMs.
Sources · arxiv.org
Study reveals AI language models develop unique biases in resume screening beyond those inherited from training data.
Sources · MIT Technology Review
Recent research reveals significant gaps in multimodal AI models' active visual perception and explores how pretraining affects reinforcement learning outcomes.
Sources · arxiv.org
