Why AI safety needs to be seen, not just stated
Every week brings another headline about AI going wrong somewhere, another set of internal questions about whether the company can really say its AI is under control. The honest answer is probably “we have controls in place, mostly, somewhere”, but with teams using AI sporadically or without being open about what they use, it’s difficult to ascertain what is really going on. There are also news articles, progress updates and new tools being released every week, it’s hard to keep up.
Buyers feel the same pressure from their side. They’re scanning, on LinkedIn, on your website, on a sales call, for whether this looks like a company that has its AI under control. For a while it felt like a game – “Can I spot where they’ve used AI?!” But now it’s becoming more serious. Underneath all of it sits a question we ask every time we hand over something important: can I see how it’s being handled? When the answer feels unclear, trust is reduced, no matter what the underlying product is actually doing.
Small details, loud signals
As a designer, I’ve spent my career paying close attention to how things look, feel, and what they imply. When a layout feels off or I spot some mis-kerned text, or inconsistent corner radii for example I wonder who approved it and why it wasn’t picked up. Or if it was even part of the brief or brand system in the first place?
Ideally, visual design shouldn’t be a job at the end of an AI build. It should be a trust-led UX challenge that is considered throughout the build process, not as an afterthought. Visible safety is the language your buyers are reading you in. If it looks consistent, flows well and is well thought out, most people will happily work their way through your product with no questions asked. If something looks inconsistent, sticky or just feels off, they start to question its provenance. We all do it with the rise of scam culture and users are already wary.
Bring everyone into the process
I would imagine that most regulated SaaS businesses already have security policies and governance that often prevent the use of AI. The challenge is how to get the best value from AI without adding any risk to their data, systems and customers.
What’s needed is the confidence and the culture to fully integrate secure systems and onboard all staff in the process. My team and I have been discussing a lot lately how it’s so beneficial to include everyone as part of the process rather than just giving a team a product and asking them to get on with it.
A clearer path to credible AI control
Many leaders are probably wondering whether they have a clear picture of how their organisation scores on AI control today. That is exactly what the AI Risk & Value Scorecard from Scail is built to answer. It’s a structured diagnostic that benchmarks a business across eight core areas of AI capability, including Governance and Control, Adoption and Integration, and Culture and Capability, three of the areas where AI transparency design and trust-led UX sit directly. The Scorecard turns a fuzzy sense of “we probably have this covered” into a clear picture of where you actually stand, where the credibility gaps are, and what a 100-day path to a stronger position looks like. AI safety isn’t just a backend issue. It’s something every regulated buyer is trying to see in your product, your brand, and the way you talk about your work. Make it easy for them to find it.
Read more about our AI Risk & Value Scorecard.
