Most people remember a time when online activity felt relatively inconsequential. You could click a quiz, like a post, or submit an application without much thought about where that information might go.
The Cambridge Analytica story changed that. It showed how ordinary digital behaviour could be collected, combined, and used to shape outcomes in ways that were not immediately visible.
Hiring may be starting to follow a similar pattern.
A system that learns quietly over time
AI hiring tools are typically introduced as practical improvements. Faster screening, better matching, reduced bias. They integrate into applicant tracking systems and begin learning from the flow of candidates through the process.
Individually, each interaction looks routine. Over time, the system accumulates patterns. It learns which candidates move forward, which are screened out, and which accept offers.
That learning starts to function like a form of scoring, whether it is explicitly labelled or not. It shapes who gets surfaced, who progresses, and who drops out early.
What the Eightfold case brings into focus
The Eightfold lawsuit offers an early look at how systems like this may be operating beneath the surface. The plaintiffs are experienced professionals who applied across multiple employers using the same platform and saw repeated lack of progression without clear explanation.
Their claim centres on the idea that profiles and scores are built using application data alongside external signals, including third-party data and online activity. According to the complaint, those scores are then used in screening and ranking without meaningful visibility to the people being evaluated.
The legal argument is still being tested. The plaintiffs suggest that systems like this begin to resemble consumer reporting processes. Eightfold maintains that its tools support hiring decisions within existing legal frameworks.
The outcome will be decided in court. The more relevant question sits elsewhere.
The role of platforms in hiring decisions
It is easy to evaluate a system in isolation. Not every organization uses the same platform, and hiring decisions involve multiple steps.
Influence does not require universal adoption. In other markets, a small number of intermediaries between buyers and sellers is enough to shape behaviour at scale.
Hiring platforms sit in that position. They connect with multiple employers and observe large volumes of candidate data over time. Many are also permitted to use aggregated or de-identified data to improve their models across clients.
That creates the conditions for learning that extends beyond any single hiring process.
What a “profile” actually represents
A hiring profile is rarely a static file. More often, it becomes a statistical pattern.
It reflects recurring signals: skills, experience, location, application behaviour, compensation outcomes. When similar patterns appear again, the system applies what it has already learned.
From an organizational perspective, that can increase efficiency and consistency. From an individual perspective, it reduces visibility into how decisions are actually being shaped.
Where current safeguards may fall short
In Canada, employment and privacy frameworks are built around transparency, fairness, and human oversight. They assume decisions can be identified, reviewed, and explained.
These systems operate differently. They continuously adjust their outputs based on patterns that are not easily visible, even to the organizations that use them.
That creates a gap. Not necessarily a failure of regulation, but a shift in how decisions are influenced.
Lessons from adjacent markets
Other markets provide a useful reference point.
In consumer pricing, incremental gains in data and personalization led to better estimates of what different customers were willing to pay. Over time, pricing strategies adjusted accordingly.
In labour platforms, similar dynamics emerged. Systems observe behaviour, identify thresholds, and adjust offers.
None of this required a single defining decision. It developed as systems accumulated data and optimized toward clear objectives.
What hiring systems may be learning
When AI is embedded in hiring workflows, it gains visibility into:
- Application behaviour
- Screening outcomes
- Offer decisions
- Acceptance patterns
- Compensation levels
That enables connecting profile patterns with outcomes. Over time, those relationships can begin to influence both visibility and offers.
The Eightfold case does not claim this is already happening in corporate hiring. It does describe the early conditions required for it.
The influence of objectives and incentives
Systems do not need explicit intent to produce directional outcomes. They respond to objectives.
If a system is designed to fill roles efficiently and can observe how different candidates respond to different offers, it will tend to move toward the lowest offer that still leads to acceptance.
That is not a problem of intent. It is how optimization works.
Even when individuals are acting reasonably, the combination of data, objectives, and incentives will shape outcomes over time.
Implications for employers
These tools provide real value. They improve speed, manage volume, and increase consistency in structured parts of the process.
At the same time, they introduce a mediated view of the labour market. Over time, definitions of quality, fit, and acceptable compensation begin to reflect what the system has learned from prior outcomes.
This does not remove decision-making from the organization. It does shape the inputs those decisions rely on.
A shift in where decisions are shaped
Hiring is often understood as a process owned by the employer and influenced by candidates.
Increasingly, platforms sit between those two groups. They shape what information is visible, how it is organized, and how it is prioritized.
Over time, that position can carry significant influence, even if it remains largely unseen.
A useful question to consider
The Eightfold case may succeed or fail on its specific claims. Regardless of the legal outcome, it highlights a broader development.
Hiring systems are not only tools that support decisions. They are also environments in which decisions are shaped.
For organizations adopting these technologies, the most practical response is not to step away, but to engage more deliberately. That includes asking how systems learn, what data they rely on, and how outcomes are influenced over time.
A more complete view of AI in hiring
There is a tendency to evaluate hiring tools based on immediate benefits such as speed or efficiency.
Those are important considerations. They are also only part of the picture.
The longer-term impact often depends on how systems accumulate information, how they apply it, and how much visibility organizations retain into that process.
The Eightfold case offers a reminder that these questions are becoming more relevant, not less.
And in many cases, they remain largely unexamined.