Alibaba’s VimRAG Scores 50.1 on Multimodal Benchmarks — Beating the Best Baseline by 6 Points
Alibaba's Tongyi Lab launched VimRAG, a multimodal RAG framework using directed acyclic graphs to improve visual reasoning and retrieval accuracy.
Archive
Category: Artificial Intelligence
Coverage of AI models, machine learning research, generative AI, and artificial intelligence industry developments
AI Was Supposed to Make Healthcare Cheaper—Instead, Costs Are Surging
Hospitals and insurers are spending billions on AI infrastructure and passing costs to patients before any efficiency gains materialize.
80 Percent of Workers Are Rebelling Against Their Company’s AI—And They Have a Point
Only nine percent of office workers trust their employer's AI tools with anything that matters. The gap between what companies promised and what workers got has…
Google’s Gemma 4 Runs a Full AI Agent on Your Phone—No Cloud, No Data Leaving Your Device
The free, open-source model processes text, images, and audio entirely on-device with built-in agent skills that autonomously use tools like Wikipedia and maps.
Anthropic’s Mythos AI Found a 27-Year-Old OpenBSD Bug That 5 Million Fuzzer Attempts Missed
Anthropic unveiled Claude Mythos Preview, an AI model that independently discovers critical security flaws across major software, outperforming traditional tools by 90x.
OpenAI Backs Illinois Bill That Would Shield AI Labs From Liability in Mass Casualty Events
OpenAI testified in support of Illinois legislation that would protect AI developers from lawsuits in cases involving mass casualties or major damages, raising concerns among safety…
Woman Sues OpenAI After ChatGPT Helped Ex-Boyfriend Create Fake Psychological Reports to Stalk Her
A California lawsuit claims ChatGPT helped a stalker produce fake clinical reports to harass his ex, raising urgent questions about AI liability and platform accountability.
Molotov Cocktail Hits Sam Altman’s Home at 3:45 AM Before Suspect Threatens to Burn Down OpenAI Headquarters
A man was arrested in San Francisco after targeting OpenAI CEO Sam Altman's residence with an incendiary device and threatening to burn down the company's headquarters,…
CoreWeave Inks $21B Deal With Meta — And the AI Infrastructure Race Just Got Expensive
CoreWeave secured a $21B agreement with Meta to deploy dedicated AI cloud capacity through 2032, with AMD issuing Meta 160M shares tied to performance targets.
Anthropic Built Its Most Powerful Cyber Weapon. Then Locked It in a Vault
Anthropic's Claude Mythos found 27-year-old vulnerabilities and thousands of zero-days -- then refused to release it publicly, launching a $100M industry consortium instead.