Daily AI News

Google Launches Agentic Video Understanding for Gemini Flash Models, Cutting Video Tokens by Up to 88%

Google introduced an agentic video understanding capability for its Gemini Flash models, enabling the model to navigate video content and load only relevant segments, reducing video token usage by up to 88%.

Read source →

OpenAI's GPT-6 Astra hallucinates less but remains vulnerable to hidden prompt injections

OpenAI's latest frontier model GPT-6 Astra reduces hallucinations compared to its predecessor and blocks 99.99% of direct prompt injections, but hidden injections within documents still succeed 8.5% of the time.

Read source →

Anthropic uses Claude to formalize proof of Fermat’s Last Theorem

Anthropic employed its Claude model to produce a computer-verifiable formal proof of Fermat’s Last Theorem, detailing the effort in a blog post.

Read source →

Nvidia bags Hugging Face in a multibillion‑dollar deal

Nvidia announced it has acquired Hugging Face, the leading open‑source AI model repository, for nearly $13 billion, aiming to integrate the platform deeper into its AI stack.

Read source →

Gimlet Labs raises $300 million for disaggregated inference platform

Gimlet Labs secured a $300 million Series B round led by Andreessen Horowitz, valuing the startup at $3 billion, to expand its inference workload acceleration platform.

Read source →

Deepseek plans massive Huawei Ascend chip cluster in Inner Mongolia

Deepseek aims to deploy 160,000 Huawei Ascend‑950DT processors in an Inner Mongolia data center for inference‑only workloads, which would be the largest known Huawei chip cluster.

Read source →

Roland launches Melody Flip generative AI music plug‑in

Roland introduced Melody Flip, a DAW plug‑in offering around 250 thematic ‘palettes’ for generative music creation, marking the company’s entry into AI‑assisted audio production.

Read source →

Lyria 3.5 music generation model now available in Gemini

Google released Lyria 3.5, its latest music generation model, within the Gemini app and API, promising more expressive vocals and richer musical arrangements.

Read source →

Google’s Gemini Spark gains ability to manage Google Photos library

Gemini Spark can now edit, curate, and create shared albums in Google Photos, turn photos into calendar events, and handle other photo‑library tasks for AI Pro and Ultra subscribers.

Read source →

Food delivery riders demand transparency of AI ‘black box’ affecting pay

Gig workers are urging delivery platforms to disclose how their algorithmic job‑allocation and pay systems operate, claiming the opacity has reduced earnings.

Read source →