Microsoft

Detecting backdoored language models at scale

Learn how Microsoft research uncovers backdoor risks in language models and introduces a practical scanner to detect tampering and strengthen AI security.

Fundamental emerges from stealth with first major foundation model trained for tabular data

Emerging from stealth, the company is debuting NEXUS, a Large Tabular Model (LTM) designed to treat business data not as a ...

Qwen3-Coder-Next offers vibe coders a powerful open source, ultra-sparse model with 10x higher throughput for repo tasks

On SWE-Bench Verified, the model achieved a score of 70.6%. This performance is notably competitive when placed alongside ...

New AI model enables native speakers and foreign learners to read undiacritized Arabic texts with greater fluency

Reading an Arabic newspaper, a book, or academic prose fluently, whether digital or in print, remains challenging for many ...

Are LTMs the next LLMs? This new type of AI can do what large-language models can’t

Fundamental, which just closed a $225 million funding round, develops "large tabular models" (LTMs) for structured data like ...
DATAQUEST

Sarvam AI launches audio model for 22 Indian languages

Sarvam AI launches Sarvam Audio, an ASR model supporting 22 Indian languages. The model handles code-mixing and accents for ...

‘Google model translated Bengali without being trained for it’: Sundar Pichai’s old clip on AI behaviour goes viral

A resurfaced clip of Sundar Pichai discussing an AI's unexpected Bengali translation ability highlights the 'black box' ...

Sarvam AI unveils 'Sarvam Vision', a multilingual document intelligence model

Sarvam AI's new Document intelligence model surpasses the actuary of Gemini 3 Pro, GPT 5.2, and other AI models when it comes ...

Roblox launches AI tech that generates functioning models with natural language

By Zaheer Kachwala Feb 4 (Reuters) - Roblox launched an artificial intelligence technology on Wednesday that can generate ...

Is your AI model secretly poisoned? 3 warning signs

Is your AI model secretly poisoned? 3 warning signs ...

Mistral's New Ultra-Fast Translation Model Gives Big AI Labs a Run for Their Money

Too many GPUs makes you lazy,” says the French startup’s vice president of science operations, as the company carves out a ...

OpenAI says new coding model helped build itself

The new coding model released Thursday afternoon, entitled GPT-5.3-Codex, builds on OpenAI’s GPT-5.2-Codex model and combines insights from the AI company’s GPT-5.2 model, which excels on non-coding ...