AI News 28th December 2025

AI News Wrap-Up: 28th December 2025

🧯 OpenAI is looking for a new Head of Preparedness

OpenAI posted a senior role centered on “preparedness” - in plain terms: spot emerging model risks before they land hard, from computer security to mental health impacts. It’s the kind of title that sounds calm and clinical… until you read what the remit really covers.

Sam Altman framed it as enabling defenders without gifting attackers sharper knives, plus thinking through how to release sensitive capabilities (including bio-related ones) and how to stay confident in systems that can self-improve. The comp listed is hefty, which reads as either reassuring or faintly ominous - maybe both.

📉 Nvidia insists it isn’t Enron, but its AI deals are testing investor faith

Nvidia’s deal-making is getting side-eye because some arrangements resemble vendor financing: investing in customers who then use that money to buy Nvidia hardware. That circular “pay me so you can pay me” vibe… it’s not illegal on its face, but it leaves people itchy.

The Guardian flags concerns around SPVs and giant, sometimes opaque partnerships - and the deeper question underneath: what happens if AI demand doesn’t grow fast enough to justify all this infrastructure and financial choreography? Nvidia pushes back hard on the comparisons, but the fact the comparisons exist at all feels… telling.

🏗️ Billion-Dollar Data Centers Are Taking Over the World

WIRED’s take: we’ve moved past “cloud” as a fuzzy metaphor and into the era of extremely physical, extremely expensive AI factories. The big players are racing to plant these compute fortresses everywhere - and each one drags a vast footprint behind it, like a comet made of megawatts and concrete.

The piece leans into the stark circularity of modern AI mega-deals and the very real downstream costs: power draw, cooling water, construction disruption, local backlash. The vibe is “this is happening whether it makes perfect sense yet or not”… which is a little scary, but also, yeah, kinda accurate.

🧩 Microsoft CEO Nadella tells managers Copilot's Gmail and Outlook integrations ‘don't really work’ and steps in to fix them

Satya Nadella reportedly tore into Copilot’s Gmail/Outlook integrations internally, calling them not that smart and, bluntly, not really working. Not the sort of feedback you frame and hang in the hallway… but it’s refreshingly direct.

What’s more interesting: Nadella’s apparently going hands-on - active in internal channels, weekly grilling sessions, even pushing consolidation moves across teams. The subtext is loud: Copilot needs to feel like a “digital worker” people depend on, not a flashy intern who sometimes forgets what day it is.

💰 AI start-ups amass record $150bn funding cushion as bubble fears mount

The Financial Times reports a huge funding surge for AI start-ups, with companies stacking cash like they’re building bunkers - “fortress balance sheets” to weather a tighter market. Investors are essentially saying: raise while the weather’s good, because the forecast might turn… or so it seems.

Mega-rounds for the biggest names are pulling the whole ecosystem upward, but the bubble anxiety sits right beside the optimism. It’s that familiar mood: “the growth is real” and “this feels frothy” in the same breath.

Yesterday's AI News: 27th December 2025

Find the Latest AI at the Official AI Assistant Store

About Us

Back to blog