Fake antivirus app TrustBastion uses Hugging Face to deliver Android malware that captures screenshots, steals PINs and shows fake login screens, according to Bitdefender.

As LLMs store sensitive prompts and metadata outside traditional controls, organizations must prepare for exposure events they can’t fully prevent.

New translation models, open speech datasets, and automatic speech recognition benchmarks aim to expand AI support for African languages.

Research reveals LLMs follow allowlist policies but systematically fail to enforce organizational prohibitions, ...

Who needs humans when a purported 1.5 million agents trade lobster memes and start their own religion? Moltbook, vibe-coded by Octane AI founder Matt Schlicht in a weekend (he cla ...

The real victory won't be in the size of the model, but in the ability to finally make it work for the person in the field.

Homegrown robotics firm Addverb has launched its first humanoid robot that is 6 ft tall, weighs 80 kg, and carries a 15 kg payload.

MIT researchers unveil a new fine-tuning method that lets enterprises consolidate their "model zoos" into a single, continuously learning agent.

A YourStory webinar brought together E2E Networks, Nvidia, and Gnani.ai to show how teams are deploying voice AI at scale, handling everything from GPU instance management to code-mixed conversations ...

On SWE-Bench Verified, the model achieved a score of 70.6%. This performance is notably competitive when placed alongside significantly larger models; it outpaces DeepSeek-V3.2, which scores 70.2%, ...

The financial cost of running LLMs is astonishing. In response, the industry has rushed toward FinOps for AI, the practice of meticulously tracking and optimizing every dollar spent on computation.

When bubbles burst, what comes next can be better, if we build it differently ...