Microsoft brings Arabic AI models to its enterprise stack

Microsoft is bringing Saudi AI firm HUMAIN's Arabic-language models into its enterprise platform. The two companies announced a long-term collaboration to put HUMAIN's ALLAM model family into Microsoft's AI ecosystem — available through Microsoft Foundry for building, customizing and deploying Arabic-language applications and agents, and wired into M365 Copilot for mission-critical productivity and business workflows. Microsoft's Forward Deployed Engineers will work alongside HUMAIN's AI experts to help organizations in the Kingdom and the wider region move from AI pilots to production-scale deployments.

This is less a product launch than a positioning move. Microsoft gets a homegrown Arabic AI partner to pair with its platform as Saudi Arabia's enterprises — and its public sector — look to deploy AI in the local context, with the security and governance requirements that come with it. Brad Smith, Microsoft's vice chair and president, framed it as "AI that understands local language and context," and HUMAIN CEO Tareq Amin teed up "the next generation of enterprise AI integration" at the LEAP tech conference. The deal also quietly signals how frontier-adjacent firms are racing to localize models rather than assume English-first AI transfers everywhere — and how Microsoft is hardening its position in an AI market Washington and Beijing are both courting. Expect the specific ALLAM capabilities in Copilot to matter more than the press release's ambitions.

A new analysis finds nearly 90% of biomedical papers published in December 2025 showed signs of AI-assisted writing. The study, reported by Nature, used an LLM-usage detector more sensitive than earlier methods — one that yields direct estimates of AI involvement rather than floor estimates. Computer scientist Dmitry Kobak of Ghent University, a co-author, said he initially doubted the numbers ("I was sure that we did something wrong") before further checks confirmed the data. The higher figure also revises prior estimates: the method puts AI use in 2024 abstracts at 31%, up from an earlier 13.5%.

The finding matters because it reframes a slow-burn worry from fringe phenomenon to baseline. If nine in ten papers in a major biomedical database show traces of AI help, the question stops being whether AI is in the research pipeline and becomes how to keep it honest — especially as models can fabricate data and peer review itself leans on AI support. It cuts in both directions: AI-assisted discovery could speed science, but readers and editors now need tools to distinguish drafting help from fabricated substance. The study is a useful reality check for anyone who still assumes AI writing is a minority habit in the lab.

With Arabic-language AI knocking on enterprise doors and AI-written research becoming the norm, where should the industry draw the line between assistance and delegation? Tell us in the comments.

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Read more

AI 101 — How do AI video generators work?

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone