Accessibility isn’t often a top priority on digital projects, even when it is a legal requirement. According to “The WebAIM Million” 2026 report, 95.9% of home pages had detected WCAG 2 failures. This report only includes failures that can be automatically detected. Colour contrast, which is probably the easiest to detect, fails in 83.9% of the sites analysed.
The percentage of failing sites has decreased 1.9% in the last six years, including a 1.1% increase from last year. It’s a painfully slow and inconsistent progress for an industry where everything moves fast. This can’t be less contradictory than it seems, though.
LLMs have taken the tech industry by storm. As with any technological innovation, there is a huge business opportunity because many companies compete in this new area, including those that are in it only for the hype.
Unlike other innovations, this is changing some organisations entirely. Not only are they measuring their employees’ performance differently, but they are also changing their hiring process.
While we are immersed in this type of transformation, everything that is not the main priority vanishes. If we are AI-first, everything else is last. Thus, the current tech environment is particularly hostile towards accessibility.
This situation has not stopped accessibility efforts. The culprit of the chronic rejection of accessibility is the culture.
The overall culture in the tech industry is surprisingly homogeneous. It’s driven by constant change and innovation at the expense of depth and maturity. Iteration is celebrated as long as the project moves forward. Iterating parts of the project to improve them without changing them is never on the roadmap.
This is great for speed. But it keeps most of the work in the shallow part of the pool. Nothing counts if it’s not easy to see or sell. We build enormous submarines that never get totally submerged. We assume they can be without trying.
The only constant improvement in accessibility has been legal. Unfortunately, the industry isn’t eager to follow the rules.
Accessibility requires time and constant work, and the culture allows for neither. There must be a conscious effort in every step of the process. This doesn’t necessarily mean more people or specific roles, but it always means having the appropriate knowledge and working on it.
Sometimes it also requires time. Iterating and observing how the project is used and which obstacles need to be removed. Like user experience, accessibility is not a guessing game but a methodical approach. Unlike user experience, accessibility is not successful if it only gets us halfway there.
Although some accessibility can be automated, some can’t. The part that can’t be automated relies on human decisions. Whether a non-textual element has meaning or not is ultimately a decision only a human can make.
However, the part that can be automated is also hurt when humans are removed from the process. How can an LLM decide if there is an accessibility issue when it’s not going to be a user, and most of the time doesn’t even have the full context? The technical aspects of accessibility are easily implemented when someone human is writing the code. We are better at considering multiple perspectives simultaneously and making nuanced decisions.
A majority of the tech industry won’t even consider people a solution. It’s easy to reduce everything to them being misanthropes. Or to conclude that they have lost the plot and are too focused on the tech part to even consider people. These are just clichés.
I would confidently say that few tech companies have a people-first approach. And even fewer that consider all people involved in the process, including their own people, with that approach. We need more of these.