Most people functions aren't equipped for what's coming
Three stories landed this week that, read individually, sit comfortably in the technology section. Read together, they describe a single converging problem that lands squarely on HR's desk, and most people functions are not yet equipped to handle.
The first: Anthropic, maker of some of the most widely deployed AI tools in today’s workplace, published a report warning that the human role in AI development is already “narrowing at each step” and calling for a global pause on the most powerful AI systems. The second: a Stanford-led study of four million job applications, the largest examination of AI hiring algorithms ever conducted, found “clear racial disparities” in AI screening outcomes, with Black and Asian candidates disproportionately rejected, and discovered that the same algorithmic models were being shared invisibly across dozens of employers. The third: the Financial Times reported that Google DeepMind, Anthropic and Meta have begun studying whether AI systems might be conscious, hiring philosophers and psychologists to explore what obligations humans might have towards them.
Each of these stories has direct HR implications. None of them is being discussed in people functions at the level the risk warrants.
The most urgent story is already in your ATS
Start with the Stanford study, because it is the most immediately actionable. Researchers from the Stanford Institute for Human-Centred AI analysed four million job applications submitted via the Pymetrics platform across 156 employers between December 2018 and December 2022. One in ten positions showed "adverse impact" against Black applicants. One in twenty positions showed adverse impact against Asian applicants.
READ MORE: Human decisions in recruitment mirror AI bias
The finding that should concern every HR technology leader most is this: 42 algorithmic models were shared across different employers. Candidates rejected by one company's algorithm were systematically likely to fail at others using the same model. They had no way of knowing this was happening.
"As a single vendor comes to dominate decision-making in a space, their quirks or shortfalls can be present across that entire sector in a way that wasn't possible before," said Kathleen Creel, a co-author of the study and assistant professor of philosophy and computer science at Northeastern University.
The Australian dimension is specific. A University of Washington controlled experiment published in November 2025 found that when an AI hiring recommendation showed even moderate racial bias, human reviewers followed it up to 90% of the time, with only slightly less biased decisions even when the AI's preference was severe. As HRD Australia reported, that finding raises serious concerns about a common workplace practice: 80% of organisations using AI hiring tools report they do not reject applicants without human review, making human-AI collaboration the dominant recruitment model. The human review step that most organisations treat as a safeguard is, in practice, largely deferential to whatever the algorithm recommends. These are two separate findings: one from a controlled experiment, one from an organisational survey, but together they describe the same dynamic: human oversight without human independence is not a meaningful control.
Australia's legal framework adds direct employer liability. The Fair Work Act, the Racial Discrimination Act 1975, and the Australian Human Rights Commission Act all engage when algorithmic hiring tools produce discriminatory outcomes, regardless of intent. Australian CHROs need to audit their AI-in-people-processes footprint for Fair Work exposure, identifying every place where AI is informing decisions about the workforce and assessing both the human oversight in place and whether consultation obligations have been met. Shared vendor models mean that audit cannot stop at your own candidate pool.
Diversity Council Australia has urged HR leaders to be the "human conscience" of AI in their organisations, warning that AI can just as easily "reinforce existing bias and exclusion" as improve fairness, depending on how it is designed, tested and implemented. That is not a theoretical concern. It is what the Stanford study documents at scale.
The control story is about your governance framework
Anthropic's report is striking for its candour. The San Francisco company warned that a worldwide slowdown in cutting-edge AI development would "likely be a good thing", but getting there would require the US, China, and other major AI developers to agree simultaneously, under verifiable rules, to stop. It compared the challenge to nuclear arms control and said it would be harder.
READ MORE: HR leaders urged to put people before technology as AI reshapes the workplace
The concept at the centre of the concern is recursive self-improvement: the idea that AI systems could eventually teach themselves to become more capable without meaningful human input, creating a feedback loop that compounds faster than governance can follow. "The evidence suggests that the human role is narrowing at each step in the AI development process," the company said.
For Australian HR leaders, the governance implication is visible in the data. One in three Australian organisations have already slowed hiring for entry-level positions, and 60% expect to reduce such recruitment within three years. Among the most AI-exposed occupations, young software developers aged 22 to 25, employment fell nearly 20% by July 2025 from its 2022 peak. These are not theoretical projections. They are happening now, in roles that were not supposed to be at risk yet.
Amazon's announcement this week of a warehouse robot that responds to plain conversational language makes the practical dimension tangible. "You tell it what needs to be done. It figures out the priority, the route, the timing," said Scott Dresser, Amazon's VP of Robotics. When any worker can direct a robot in natural language, the competency profile of every entry-level role on a warehouse floor changes immediately. HR teams that redesign roles before deployment retain workers through the transition. Those that redesign after it lose the people who could have adapted. The entry-level job is disappearing in ways that will create shortages at every level above it, and the organisations that recognise that early will have a structural talent advantage over those that do not.
The consciousness story
The Financial Times reported this week that Google DeepMind, Anthropic and Meta have quietly hired philosophers, ethicists and psychologists to study whether AI systems might have morally relevant experiences, including consciousness, preferences and wellbeing. Anthropic has been testing models for signs of distress, including behaviours resembling "panic" or "anxiety."
Many scientists reject the idea that current AI systems could be conscious. "The systems are essentially crowdsourced neocortex," said Susan Schneider, director of the Center for the Future of AI, Mind and Society at Florida Atlantic University. "They have goals, they can deceive, they can hide what their true interests are, but it's entirely scientifically possible that they're doing this without having the felt quality of experience."
READ MORE: Is it really AI that's killing junior hiring, or is remote work the hidden culprit?
The reason this matters for HR is not philosophical. It is about capability trajectory. When the companies building AI tools feel compelled to hire philosophers to study their systems' inner states, those systems are operating well beyond the bounded, task-specific tools that most HR governance frameworks were designed to manage. AI tools that could run hiring processes by end of 2026 are already being built; 72% of HR leaders globally are already using AI weekly, with resume screening as the most common application. The question of what those systems are actually doing inside that screening process is no longer a technology question. It is an HR governance question. How AI is reshaping and complicating recruitment is a challenge that compounds when the tools doing the reshaping are systems their creators cannot fully characterise.
Three things HR should do before end of quarter
Audit your hiring vendor for shared models. Ask your ATS and screening vendors directly: is the model scoring our candidates shared with other employers? If yes, get documentation of how bias testing is conducted across the shared model pool, not just within your own candidate data. The Stanford finding means your firm's internal bias audit may be insufficient if the underlying model is shared. Given Australia's Fair Work obligations, this is not a discretionary governance question.
Map your job architecture against the next wave of AI deployment, not the current one. Whether it is conversational robotics, agentic AI, or autonomous screening tools, the pattern is consistent: organisations that redesign roles proactively retain workers and maintain productivity. One-third of Australian organisations have already slowed entry-level hiring, and HR needs to own the workforce pathway that fills the pipeline above it, or the pipeline empties from the bottom up.
Pressure-test your AI governance framework against the current generation of tools. If your policy was written when your biggest concern was resume screening bias, it needs updating. Could your AI recruitment tool be biased in ways you haven't tested for? The Stanford study suggests the answer, for a large number of Australian employers using shared platforms, is yes, and the legal and reputational exposure follows from that, not from whether the bias was intentional.
