Tech

Meta strikes AI licensing deals with CNN, Fox News, and USA Today

Posts from this topic will be added to your daily email digest and your homepage feed.

Waymo’s robotaxis are under investigation for passing stopped school buses

Posts from this topic will be added to your daily email digest and your homepage feed.

EU fines X $140 million over ‘deceptive’ blue checkmarks

Posts from this topic will be added to your daily email digest and your homepage feed.

Chicago Tribune sues Perplexity

The newspaper is alleging copyright infringement and calling out Perplexity's Retrieval Augmented Generation (RAG) as a culprit.

Gemini 3 Deep Think is rolling out now.

The enhanced reasoning mode is only available to Google AI Ultra subscribers through the Gemini app. It’s currently the highest-performing model on the ARC-AGI-2 reasoning benchmark, with Google saying it’s “designed to tackle complex math, science and logic problems that challenge even the most advanced state-of-the-art models.”

Micro1, a Scale AI competitor, touts crossing $100M ARR

Micro1 started the year with roughly $7 million ARR. Now, it claims to have surpassed $100 million in ARR, double what it reported in September.

Anthropic CEO weighs in on AI bubble talk and risk-taking among competitors

Anthropic's CEO shared his thoughts on the economics of AI and the risk-taking of competitors, saying some were "YOLO-ing" with regard to spending.

Researchers find what makes AI chatbots politically persuasive

A massive study of political persuasion shows AIs have, at best, a weak effect.

Why won’t Steam Machine support HDMI 2.1? Digging in on the display standard drama.

Valve tells Ars its “trying to unblock” limits caused by open source driver issues.

Microsoft is quietly walking back its diversity efforts

Posts from this topic will be added to your daily email digest and your homepage feed.

Jina-VLM: Small Multilingual Vision Language Model

We're releasing jina-vlm, a 2.4B parameter vision-language model that achieves state-of-the-art multilingual visual question answering among open 2B-scale VLMs. By combining a SigLIP2 vision encoder with a Qwen3 language backbone through an attention-pooling connector, jina-vlm delivers strong performance across 29 languages while remaining efficient enough to run on consumer hardware.

Google’s AI model is getting really good at spoofing phone photos

Posts from this topic will be added to your daily email digest and your homepage feed.