Duty of Care
Looking out for humans in the Age of AI.
Programming note:
Hello to my +50 new subscribers! Who knew a comment about AI slop on Substack would pique such interest! Welcome and hope you enjoy my Stack. And apologies to everyone for a few weeks’ hiatus due to planned medical leave (don’t worry, I’m fine!). I will be back in 2026 with my (hopefully) 3 - 4 posts a month.
What a wild ride 2025 has been in the world of AI (and in the world, generally).
We’ve had continuous drops of updated frontier models, with capabilities expanding at a relentless pace. We now have AI-driven image and video generation that has undermined our ability to determine what’s real and what’s fake. There have been incredible improvements in autonomous Agents, which can now complete full workflows on our behalf independently, at least on some things.
And we have had a steady flow of doomsday headlines telling us that AI is coming for all of our jobs. These threats are often released on a ‘brocast’, originating from a ‘godfather of AI’ or a tech billionaire telling us AI will eliminate knowledge work as we know it, in anywhere from a few years to a few decades.
While I try to operate from a place of hope and abundance, make no mistake - I wholeheartedly believe the risks from AI are very real: to our attention, to our jobs and to the very fabric of how society operates.
But it’s important to remember that this dark future being fed to us is not inevitable. The current narrative around AI is designed to make us believe we are locked onto a path to a future unknown, and we must accept the multitude of potential risks for the sake of ‘progress’. CEOs of big AI companies acknowledge the challenges in front of us, yet still pursue model capability growth aggressively because someone else ‘will get there first’ [mostly China]. Put simply, the West has made most people believe AI is a zero-sum game where America’s model makers must win, or we are all doomed.
But in all the coverage and debate around AI, there is one thing that is noticeably absent from the conversation: that our future is a choice. Speaking about AI like it’s a foregone conclusion has made us forget that we can influence and shape how we use, regulate, deploy, and develop this technology. In reality, I think we all have more power than we realise.
We all have the agency to make a difference when it comes to AI. I believe that we can mitigate some of the known risks with AI - individually, in our companies, and at the societal level. Of course, the bigger the system, the harder it is to manage and change, but that doesn’t mean we shouldn’t try.
In the workplace, each and every one of us can decide to take concrete action towards a different future. Whether you are in a position of power that impacts or an individual who is curious about this technology - we all can do something.
But how? How can we actually create a future that makes the most of both Human and AI capabilities? How can we protect humans - their brains, their unique abilities and their attention - from AI? And what can you do tangibly to make a difference today?
Unsurprisingly, I have ideas. They all coalesce around a simple concept that is applicable if you are thinking about using AI individually, on your teams or in your organisations. When it comes to AI, we all should be thinking about our duty of care.
Duty of care in the Age of AI
When I say duty of care, you are likely thinking of those in positions where they have legal and ethical obligation to act with reasonable care to avoid causing harm to others. Think doctors and the Hippocratic Oath. Or police in their obligation to protect society writ large, only using appropriate force. Even investment bankers, who have legal requirements to operate within a defined framework designed to prevent harm.
When it comes to organisations, the standard duty of care can be understood as employers and leaders acting responsibly to protect their people from foreseeable risks, whether physical, psychological, or economic. And I argue, when it comes to AI and work, duty of care should take on new dimensions.
We’re asking people to adopt tools that can fundamentally reshape how they think, work, and create value. Tools that come with known risks around hallucinations, bias, data privacy, cognitive offloading, and job displacement. And yet most big companies are racing to implement AI without the same rigour they’d apply to any other significant workplace change. We wouldn’t hand someone heavy machinery without proper training and safety protocols. Why should AI be any different?
Duty of care in the Age of AI extends beyond compliance, security and risk mitigation. It extends to the choices we make today about how we deploy, govern, and use this technology will shape the future of work for everyone, at every level.
So what does a duty of care look like when it comes to AI, specifically, at this moment?
For individuals, you’re the only one who is responsible for your own AI use (or non-use). And for yourself, a duty of care looks like:
Building your own AI fluency, which includes understanding the known risks of using generative technology, including but not limited to hallucinations, bias, data privacy, cyber security.
Applying your judgment and discernment when using AI, which includes fact-checking references, using your critical thinking to evaluate generated outputs, and evaluating when and how it’s appropriate to use AI.
Benchmarking AI capabilities against your own workflow standards on an ongoing basis, evaluating what’s working and what’s not as models evolve.
Intentionally protecting your brain and attention from the known risks related to AI use, including cognitive offloading.
Team leaders and managers have an expanded duty of care, where it starts to cross into organisational expectations. Duty of care at this level looks like:
Modelling exemplary AI use (see above).
Giving your team the time and space to build their AI literacy, experiment with AI-available tools on your team and collectively make sense of AI use (i.e. - what’s working and what’s not).
Designing and implementing working practices that enable appropriate AI use and set the foundations for leveraging AI capabilities in future.
Guiding your teams through intentionally designing workflows and use cases that makes the most of both human and AI capabilities.
Setting clear expectations and boundaries about using AI and holding people accountable for inappropriate AI use.
The C-suite and senior leaders have a significant duty of care, to their shareholders and to employees. In the Age of AI, managing the tensions can be difficult, and it adds significant complexity to how they operate. Leaders need to invest in AI while ensuring business-as-usual delivery, all while avoiding costly mistakes with a relatively new technology and transforming their workforce ethically and sustainably. Duty of care at this level looks like:
Strategically investing in a few AI ‘big bets’ that are appropriate for your business. At this stage, these are likely highly specialised, highly targeted use cases that are rigorously built on valid data sets that can truly change how you deliver value.
Creating internal capabilities that will protect both the company and humans using AI, such as fair AI use policies, ERC requirements, AI governance frameworks, and multi-faceted support for your people (see below on HR’s duty of care).
Providing access to a reasonable GPT in your enterprise tech stack, inclusive of multi-modal capability.
Recognising that this is as much an operating model issue as it is a technology one: you will not get returns on AI investments without thinking about and invest into the people using it.
Refusing to set or implement headcount reduction targets associated with AI implementations. There will be productivity gains and the workforce will change and shift over time: however, focusing solely on job losses only yields short-term benefits. Repurposing productivity for growth and innovation is the far more sustainable play.
Investing in reskilling and upskilling your people. The WEF Future of Jobs report states that 59% of people will need some form of skills development by 2030. The skills you’ll likely need will likely be in short supply in the labour market, so you need to build them internally, and you need to start thinking about this now.
Taken together, these lists aren’t exhaustive, but they are a starting point for how we can think about a duty of care in AI at every level of an organisation. But there is one function above all others that should be leading the human side of the AI transformation and the future of work: Human Resources.
HR’s Duty of Care in the Age of AI
HR - in many places - has a bad reputation. And to be fair, some of it is warranted. They claim to be supporting employees, but when things get tough, they side with the company. Mistakes HR makes can cost people their jobs, their mental health, and in some horrible cases, their lives.
But having worked for many years alongside HR, I can attest that many people in the function are trying to do the right thing. They deeply care about employees, they stand up to leaders making poor (and sometimes illegal) choices and they are deeply concerned about the impacts of AI on their workforces.
Currently, HR is being asked to manage the human consequences of AI-related decisions made without them. But this doesn’t relieve HR of their responsibilities, and HR needs to step up and be honest about what’s happening. They need to take ownership and level up their duty of care.
Given the pace of change, I think it’s useful to think about HR’s evolving responsibilities across three time horizons.
Horizon 1: Now – 3 to 5 years
In the near term, HR must build the foundations that protect both their people and their organisations, which looks like:
Building your own AI fluency first, as you cannot guide this transition if you don’t understand the technology yourself.
Advocating loudly for the people side of AI transformation, early and even when it slows things down.
Creating space and resources for employees to build AI literacy, experiment, and collectively make sense of how AI fits into their work.
Developing fair AI use policies and governance frameworks that protect both the company and the humans using AI, especially around discrimination, bias and fair use.
Refusing to be complicit in poorly designed workforce reductions justified by AI. Challenge headcount targets that aren’t grounded in genuine workflow redesign.
Starting task-level job analysis now to understand which tasks are most susceptible to AI augmentation.
Horizon 2: 3 – 7 years from now
As AI diffuses more widely and job losses increase, HR’s duty of care expands significantly to include:
Designing and leading genuine workforce transformation through reskilling and redeployment, not just headcount management.
Redesigning jobs intentionally around the evolving division of labour between humans and AI.
Protecting employees proactively from the cognitive and psychological risks of heavy AI augmentation, including cognitive offloading.
Managing tensions between workers who excel with AI, those struggling to adapt, and the emerging AI Agent workforce.
Building skills-based talent approaches that enable rapid redeployment as work and organisations continue to shift and evolve.
Preparing for the regulatory changes that will inevitably come around AI-assisted decisions and workforce surveillance.
Horizon 3: 7 – 10 years (or more) from now
In the longer term, the fundamental assumptions underpinning HR start to collapse. We don’t know what work will look like, but there are strong signals that it will be radically different what it is today. As a result, HR’s duty of care will include:
Rethinking every traditional HR domain – reward, performance, talent management, L&D, org design – from first principles.
Championing new approaches to valuing human contribution when output is no longer tied to hours worked and knowledge is no longer a unique human capability.
Designing for a different employment ecosystem where AI-augmented talent moves fluidly between teams, organisations and labour markets.
Advocating for broader societal changes – labour law reform, portable benefits, new safety nets – to support workers through this transition.
Keeping the human at the centre, even as the definition of ‘work’ transforms beyond current recognition.
HR has largely been absent from the AI conversation at the strategic level. They’ve been handed implementations to execute rather than invited to shape the direction. That needs to change – not because HR deserves a seat at the table, but because the transition ahead is fundamentally a human one. Get this wrong, and HR will have enabled harm at a scale we’ve never seen. Get it right, and the function becomes essential to navigating one of the most significant transitions in the history of work.
Who is looking out for the humans?
Easy to answer. In the context of work, my view is clear: we all have a responsibility to protect ourselves, our colleagues and our organisations in the Age of AI. And the greater your power in a company, the greater the need to provide a duty of care to the humans in it.
I know that some days it’s hard to hold onto hope. The headlines are relentless. The pace is disorienting. And the people with the most power to shape this transition often seem the least interested in the human consequences.
But keep coming back to my own agency. That I have the ability to shape and influence and advocate for a future I want to be a part of. It’s important that every interaction you have about AI has the potential to be a change intervention. Every conversation about AI with a colleague, every AI-related policy decision, every choice about how to use this technology in your own work all matter. These moments add up and they shape what’s possible.
So as you close out this year - one of WILD transition and AI progress - it’s so important to remember that we all have more agency than we think. The future isn’t being done to us. So if you’re reading this and wondering what you can do, start with the list above.
Because no one is coming to save us. We have to look out for ourselves, and each other.
Inspiration for my post came from an article written last year by my colleague Meir Adler on HR’s Duty of Care: Read it here.