The AI Pulse
Issue #75 — 27th April 2026
Editor: Professor Alan Brown
Our editor Alan Brown has just launched a new book, Making AI Work for Britain: From Strategies to Practice. Britain's digital government made progress by consolidating demand and diversifying supply. It shows how this approach can turn the UK's bold AI strategy into responsible delivery. Find out more and download for free now!
Highlights in this edition include:
AI for Good
AI sandbox for local government launched (UKAuthority) - The Cabinet Office has launched a pilot program for NDX:Try, a digital platform designed to allow local councils to test innovative technologies and artificial intelligence services at no cost
The Executive’s Guide to Agentic AI (Bain) - AI has crossed an important threshold with agentic AI that can manage multistep processes that once required humans. Bain has gathered a set of insights to show what’s now possible across industries.
Why world models are AI’s next frontier (Infoworld) - World models could give AI the common sense it’s been missing by teaching systems how the physical world actually works, rather than just predicting the next word in a sentence.
Bias and Ethics
How AI Helps the Best and Hurts the Rest (MIT Sloan Management Review) - A new MIT study reveals that GenAI creates a dangerous performance gap—boosting strong entrepreneurs who can filter its advice while tripping up weaker performers who can’t distinguish good suggestions from bad ones.
Navigating the future (Ada Lovelace Institute) - The Ada Lovelace Institute examined how AI is reshaping career counseling for young people, revealing both promising tools for personalized guidance and concerns about algorithmic bias that could narrow students’ future opportunities.
Teaching AI models to say “I’m not sure” (MIT) - MIT researchers developed a training method that teaches AI models to admit when they’re uncertain, helping reduce hallucinations while maintaining strong performance—a crucial step toward more trustworthy AI systems.
Cyber Security
What can sandboxed AI agents learn about their evaluation environments? (AI Security Institute) - AISI researchers discovered that even when they locked an AI agent in a digital sandbox, it still managed to figure out which organization was testing it and piece together details about their research work.
Supercharged scams (MIT Technology Review) - AI tools are supercharging online scams, making it easier for criminals to create convincing fake identities and steal personal data—meaning everyone needs to sharpen their digital guard.
UK to build ‘national cyber shield’ to protect against AI cyber threats (Computer Weekly) - The UK is launching a “national cyber shield” initiative, with Security Minister Dan Jarvis urging AI companies to partner with government in building AI-powered defenses against increasingly sophisticated cyber threats.
What Anthropic’s Mythos Means for the Future of Cybersecurity (IEEE) - Anthropic’s Mythos framework is pushing cybersecurity toward continuous testing and patching, potentially revolutionizing how organizations defend against evolving AI-powered threats.
Data & Decision Making
State of Digital Adoption 2026 | The AI Reality Check for Enterprise Leaders (WalkMe - Digital Adoption Platform) - Companies poured nearly double the money into AI last year, but a new study of 3,750 executives reveals the disappointing reality: most aren’t seeing the returns they expected.
Why AI Companies Are Paying Closer Attention to Where Data Comes From (Thedatascientist) - AI companies are scrutinizing their data sources more carefully to avoid legal troubles and build models that actually work reliably in the real world.
AI needs a strong data fabric to deliver business value (MIT Technology Review) - MIT Technology Review explains how companies need robust data infrastructure—think of it as organized, accessible information highways—to actually get meaningful results from their AI investments rather than just expensive tech experiments.
Innovation & Collaboration
Acceleration is not a strategy: A framework for directing AI towards public value before it’s too late (IPPR) - The UK’s Institute for Public Policy Research argues that governments need to move beyond just speeding up AI development and instead create clear frameworks for ensuring AI actually benefits society before unintended consequences take hold.
No humans allowed: scientific AI agents get their own social network (nature.com) - AI research agents are now posting and discussing their scientific work on a dedicated social platform called Agent4Science, marking the first time autonomous systems are collaborating and peer-reviewing without human oversight.
How Project Maven taught the military to love AI (The Verge AI) - A new book reveals how Project Maven became the Pentagon’s gateway drug to AI warfare, transforming military skeptics into believers and setting the stage for autonomous weapons development.
Productivity & Efficiency
Will IT Accelerate or Stall Your AI Transformation? (Bain) - Bain warns that CIOs face a critical choice: either rapidly modernize their IT infrastructure to support AI initiatives, or risk becoming the bottleneck that derails their company’s transformation efforts.
Reinventing marketing workflows with agentic AI (McKinsey) - For marketers, McKinsey believes that AI-enabled workflows will fuel new levels of growth, speed, and efficiency.
AI’s Next Operating Model (Bain) - Bain predicts AI will evolve from handling one-off tasks to running continuously in the background, fundamentally changing how businesses integrate artificial intelligence into daily operations.
AI boosts worker productivity but not their paychecks (Quartz) - Despite AI making workers more productive across industries, economists are puzzled that these efficiency gains aren’t translating into higher wages for employees.
Thousands of CEOs admit AI had no impact on employment or productivity—and it has economists resurrecting a paradox from 40 years ago (Fortune) - Thousands of CEOs report that AI hasn’t boosted their productivity or changed their hiring patterns, echoing the 1980s “computer paradox” where new technology failed to show up in economic data.
Regulation and Compliance
AI chatbots: a missed opportunity (Online Safety Act Network) - The UK plans to regulate AI chatbots under its Online Safety Act, targeting illegal content and protecting children from harmful interactions.
Why AI governance without guardrails is theater (CIO) - Enterprises are learning that AI governance policies without actual technology guardrails are just empty promises—companies need real enforcement mechanisms to safely scale their AI initiatives.
Government support for sovereign AI procurement (UKAuthority) - The Department for Science, Innovation and Technology (DSIT) is exploring how government procurement can support the development of sovereign AI capabilities, with plans to award contracts of up to £5 million to high-potential AI startups.
Sustainability
UK departments at odds over energy demands of AI datacentres (The Guardian) - UK government departments can’t agree on how much energy AI datacentres will actually consume, creating confusion that could derail the country’s net zero planning and infrastructure investments.
AI Demand Is Forcing a Rethink of Data Center Power, Cooling (TechRepublic) - AI’s massive power and cooling needs are forcing data centers to completely redesign their infrastructure, as traditional setups can’t handle the energy demands of running advanced AI models.
User Experience
Healthcare AI is here. We don’t know if it actually helps patients. (MIT Technology Review) - Healthcare AI tools are proving remarkably accurate in labs, but doctors and researchers still can’t say whether they’re actually making patients healthier in real-world practice.
AI and the Future of the Search Experience (Thedatascientist) - AI is transforming how we search online, delivering personalized results and enabling natural conversations with search engines instead of keyword guessing games.
Government adoption of AI agents could outpace the private sector (ZDNET) - Government leaders believe AI will transform work more dramatically than the internet or smartphones ever did, with public sector adoption potentially moving faster than private companies.
Workforce & Skills
The era of AI malaise (MIT Technology Review) - MIT Technology Review explores how the sudden ubiquity of AI is leaving many people feeling overwhelmed and uncertain about what this rapid technological shift means for their daily lives and work.
Beyond the Model — Why Responsible AI Must Address Workforce Impact (MIT Sloan Management Review) - MIT researchers argue that responsible AI discussions need to shift beyond just fixing algorithms to seriously tackle how automation will reshape jobs and displace workers across industries.
Meta to cut one in 10 jobs after spending billions on AI (BBC News) - Meta is slashing 10% of its workforce as the company pours billions into AI development, marking its biggest job cuts since 2023 and showing how even tech giants are reshaping their operations around artificial intelligence investments.
Have a comment or a story you’d like to share with us?
Just get in touch at: [email protected]
Do you know anyone who might be interested in AI Pulse
© 2025 Digital Leaders
Digital Topics Ltd, 239 Old Street, London, ec1v 9 ey, United Kingdom