How much of a problem is AI’s water use?
AI’s water consumption is a growing concern, but its impact depends heavily on location and cooling technology. Data centers use relatively little water nationally, yet can significantly strain supplies in drought-prone regions.
Editor's noteHow much water AI uses depends almost entirely on where the data centre sits and how it is cooled, and that is why it led Technology: nationally the number is small, in the wrong county it is everything. An average hides a drought. If you site compute or buy it, the water question is a location question, answered by postcode.
TECHNOLOGY
Bill Gates Says Some Jobs Should Be Off-Limits to AI
Bill Gates argues that some roles should remain permanently human, even if AI can perform them. He proposes “Human Reserved” jobs and taxes on AI and robots to reduce worker displacement and preserve social value.
Crescent Island: Turning Memory Capacity into Agentic AI Throughput
Intel’s Crescent Island is an AI inference GPU designed for agentic AI workloads. Rather than maximizing raw compute, it emphasizes massive memory capacity, efficiency, and throughput, enabling larger models, longer contexts, more concurrent AI agents.
From Spec To Formal Properties
LLMs are increasingly capable of converting natural-language specifications into formal verification properties, but success still depends on specification quality, human oversight, completeness checks, and careful validation of AI-generated outputs.
MARKETING & DESIGN
Meta addresses ‘pervert glasses’ reputation with a privacy fix and a new marketing campaign
Meta has closed a privacy loophole in its AI smart glasses that enabled covert recording by covering the recording indicator light after capture began. The fix is accompanied by a public-awareness campaign to rebuild trust.
Editor's noteMeta closed the loophole that let its glasses keep recording with the light covered, and shipped the fix with a marketing campaign. That pairing is why it led Marketing & Design: the patch is engineering, the campaign is the admission that the reputation problem outlived the bug. When a product’s nickname is the story, a silent fix confirms the nickname.
The Future of Trust: Where AI Meets Human Connection
As AI becomes embedded in everyday experiences, trust will increasingly depend on combining technological efficiency with authentic human relationships. Brands that balance convenience with empathy, transparency, and expertise are best positioned to win consumer confidence.
Instagram will now edit the first draft of your Reel for you
Instagram has launched First Draft, an AI-assisted Reel editing feature that automatically trims clips, removes pauses, and assembles an initial cut in seconds. The tool aims to reduce editing effort while keeping creators in control.
ADVERTISING & PR
Spotify Takes Playlists to the Streets in Reactive Uncommon OOH
Spotify’s new“Playlist It” campaign turns user-created playlist titles into hyperlocal billboards across New York, Los Angeles, and Miami, celebrating playlists as a form of self-expression and connecting music culture to specific places and moments.
Editor's noteSpotify put users’ own playlist titles on billboards in the neighbourhoods they were made for, and it led Advertising & PR because the creative was already written by the audience. Hyperlocal out-of-home with copy the customer wrote is the whole brief done in one move. For OOH planners, relevance is cheapest when you borrow it from the people walking past.
Apple Maps Has Ads Now. They Might Not Be Terrible, but You Can’t Turn Them Off
Apple Maps is now showing sponsored business listings in the US and Canada. While the ads are relatively unobtrusive and privacy-focused, users cannot disable them, marking the end of Apple Maps’ longstanding ad-free experience.
Dentsu expands and consolidates its commerce and retail media offering under new leadership
Dentsu reorganized its commerce and retail media unit under new brand-side hires, consolidating shopper marketing and media into one offering as US commerce media spend grew 25% year over year.
HUMAN RESOURCES
Tying bonuses to badge swipes is how you teach people to game you
Linking compensation to office badge swipes is a classic management mistake: employees optimize for the measured activity rather than the desired outcome. When attendance becomes the target, people learn to game attendance.
Editor's noteTie bonuses to badge swipes and you get exactly what you measured: swipes. Coffee badging is the workforce doing the maths it was asked to do. It led Human Resources because attendance is this year’s most visible case of a metric becoming a target and ceasing to measure anything. Before rewarding a number, ask what behaviour that number cannot see.
AI resume screening is creating a ‘visibility gap’ for recruiters
A growing paradox in AI recruiting: automation helps recruiters process overwhelming application volumes, yet many fear qualified candidates are being filtered out before a human ever reviews them, creating a troubling “visibility gap.”
Abstinence is not an AI strategy
Banning or avoiding AI in the workplace is not a viable strategy. Learning leaders should focus on responsible experimentation, AI literacy, governance, and workforce adaptation rather than attempting to prevent employees from using AI.
PUBLIC POLICY
NATO’s Missing Link: Shielding Vulnerable Infrastructure
NATO’s infrastructure investments need alliance-wide resilience assessment, not just national review. Critical assets such as energy grids, railways, data centers, and logistics networks should be evaluated for redundancy, survivability, and strategic military value.
Editor's noteGrids, railways and data centres are now military assets whether or not their owners think so. We ran this because it argues for assessing them alliance-wide instead of nation by nation, which is the difference between a plan and a wish. The private operators who own most of that infrastructure should read it as a preview of obligations.
Alternative Visions of the Internet
Today’s internet was shaped by policy choices, not technological inevitability. Reflecting on Rebecca Coyne’s spam-regulation analysis, contributors contend that stronger democratic governance and public participation can still redirect AI and internet development.
Singapore to Require Stronger Social Media Age Checks for Young Users
Singapore plans to require stronger age-assurance measures on social media platforms, moving beyond self-declared ages. The government is considering age-based access rules and may impose stricter restrictions on platforms that fail to protect younger users.
