The Glass Box Principle
What it is and how to write yours
Most AI notices in recruitment do one of two things: say nothing useful, or say so much that nobody bothers to read them. The Glass Box is the alternative; the principle that transparency about your AI is not just a compliance obligation, but a genuinely better way to treat candidates.
%20(1)-1.png?width=1200&height=800&name=AI%20Landing%20page%201%20x%20800%20px)%20(1)-1.png "AI Landing page 1 x 800 px) (1)-1")
Most firms approach AI compliance as a legal risk to minimise. Draft the notice, tick the box, move on. The minimum viable disclosure, the vaguest language that still technically counts. That approach produces what we call frosted glass: technically there, completely opaque. "We may use AI tools in our process" is frosted glass. It says nothing, discloses nothing, and builds no trust.
A Glass Box is the opposite. It's your commitment that if a candidate or employee asks "what is that AI doing with my information and why?", you can actually show them. Not because the law requires it (though increasingly it does) but because it's the right way to treat people, and because candidates and clients are increasingly asking the question.
Let's start here...
What is the Glass Box Principle?
Let's start with the name, because it's doing a lot of the work for us.
A black box is an AI system nobody can see inside. Inputs go in, outputs come out, and nobody really knows why. A glass box is the opposite; the process is visible, the logic is understandable, and the people affected can see what's happening to them. It doesn't mean explaining every line of code. It means being able to answer the question a candidate is actually asking: "What is your AI doing with my information, and why does it matter?"
The reason this has gone from being nice-to-have to necessary is that increasingly, the law is asking for it. Illinois HB 3773, in force since January 2026, requires plain-language disclosure of which employment decisions AI influences, what data it uses, and where candidates can go with questions. The EU AI Act requires transparency to affected individuals as a high-risk AI obligation. The ICO in the UK has published AI transparency guidance that, while not yet mandatory, is increasingly treated as the standard of care.
But here's the thing. Even before things got legal, recruiters that could articulate their AI use clearly had a competitive advantage. In a market where candidates are increasingly wary of algorithmic processing, the company that says "here's exactly what our AI does, here's a contact if you have questions, and here's how you can opt for a different assessment" is the one that not only looks like it knows what it's doing but the one that looks like it really cares. That trust matters, with candidates and with clients.
What a Glass Box statement actually contains
There are five things a complete Glass Box statement addresses.
Not all five need to appear in every piece of candidate-facing copy so you don't have to put the full version in every job advert, but they should all be covered somewhere in your candidate journey, and available to anyone who asks. ➡️
.png)
Let's break those down...
1. What the AI does
Describe the tool's function in plain terms. Not "machine learning system." Not "algorithmic processing." "We use an AI tool to summarise your CV and identify skills that match the role requirements." One sentence. Understandable to anyone.
2. Which decisions it influences
Be specific about the role AI plays. Does it affect who gets shortlisted? Who gets contacted first? Who is presented to the hiring manager? Say exactly that. "This tool influences which applications our recruiters review." If AI makes no final decision, say so clearly.
3. What data it uses
Name the data categories briefly. Categories only you don't need to list every field. "Your CV, covering letter and interview audio" is correct. "All available data" is not. Be specific enough that a candidate understands what they've shared that's being processed.
4. Who makes the final decision
Human or AI? B be clear about this. If a human makes the final call, say so: "A recruiter makes all final decisions." If an AI determines the outcome, be honest about that too. Candidates have a right to know — and under UK GDPR Article 22, a right to request human review of solely automated decisions.
5. What to do about it
This is the element most firms miss. Provide a named contact or email for AI-related questions. Offer candidates the right to request an alternative assessment method. Not all jurisdictions require this yet — but all are moving toward it, and building it in now costs almost nothing. It also signals to candidates that you're genuinely confident in your process.
"We use technology-assisted tools as part of our modern hiring approach. We are committed to equal opportunities for all candidates."
"We use an AI tool to review and summarise your CV against the requirements of this role. This influences which candidates our recruiters review first. The AI does not make the final decision. Questions? Contact [name/email]."
The practical bit...
Where your Glass Box statement should appear
The statement isn't a document that lives on a compliance shelf. It's a live, candidate-facing communication that needs to appear at the right moment in the right place. Illinois HB 3773 requires disclosure to candidates before or at the time AI is used, which in practice means before the AI has touched their application.Job postings / adverts
A brief version (2–3 sentences) explaining that AI is used in the application review process and where to go with questions. This covers the "before submission" requirement.
Application portal confirmation page
When a candidate submits an application, show a clear notice before they hit send.This is the clearest "at the point AI will be used" moment.
Interview invitation
If AI tools are used in interview analysis (call transcription, video analysis), a specific notice must be provided before that tool is used — not bundled into a general recruitment privacy notice.
Initial candidate communication
Your first automated email to a new applicant is a natural place to include a slightly fuller version, with contact details for questions.
Annual notice to employees
A dedicated annual communication, not buried in a broader HR update, covering which AI tools are used in employment decisions affecting them.
Within 30 days of new AI tool
Triggered whenever you adopt a new tool or a vendor makes a significant update. This one needs a process behind it.
The 30-day update trigger — and how to manage it
The requirement to update notices within 30 days of a new or substantially changed AI system is the operationally awkward part of Illinois HB 3773. Your vendor updating their model even silently, even with no change to the product name or price can trigger this obligation. You need a process for knowing when it happens.
Practically, this means: subscribing to every AI vendor's release notes and model update notifications. Assign someone to review them monthly. When a substantial update occurs, refresh your Glass Box statement and push updated notices through your candidate communication workflow.
The do's and dont's...
The easy guide to what to do - and what not to do
%20(1).png "AI Landing page 1 x 800 px) (1)")
%20(2).png "AI Landing page 1 x 800 px) (2)")
Review and revise...
Keeping your Glass Box current - don't set and forget
The single most common mistake firms will make with Glass Box statements is writing a good one and then leaving it alone. The statement is only accurate as long as your AI stack is unchanged. When a vendor updates their model, when you add a new tool, when you change how a tool is used in your workflow — the statement needs updating too.
The Glass Box Principle isn't a project with a completion date. It's an ongoing practice. The firms that get it right treat it the same way they treat any other quality assurance process: document it, assign ownership, build it into your calendar, and audit it regularly.
Here's a practical process to help keep it current:
-1.png?width=1000&height=667&name=AI%20Landing%20page%201%20x%20800%20px)-1.png "AI Landing page 1 x 800 px)-1")
Getting you started...
Some simple Glass Box statement templates
We thought it might be useful to give you some Glass Box statement templates to get you started. We've picked 3 of the Bullhorn recruitment workflows where AI tools are already in common use and a bonus one for an annual update to your existing employees or workers. Copy, paste and adapt to your specific circumstances, run it by your legal team if needed, then you'll be ready to go. *
Template 1 - CV Screening AI
How we use AI in reviewing your application
At [Company Name], we use an AI-powered tool to help us review applications for [role type / all roles]. Here's what that means in practice.
What the AI does: The tool reads and summarises your CV and covering letter, identifying skills, experience and qualifications relevant to the role. It may generate a relevance score or ranking to help our recruiters prioritise their review.
Which decisions it influences: The AI output influences which applications our recruiting team reviews first. It does not make final hiring decisions. All decisions about who to invite for interview or progress through the process are made by a qualified recruiter.
What information it processes: Your name, contact details, work history, education, skills and any other information included in your CV or application form.
What it does not do: The tool does not assess your photograph, age, location, or any information that could be used as a proxy for a protected characteristic. [If using Warden AI or similar: It is continuously monitored for fairness by an independent auditing tool.]
Your rights and options: If you have questions about how AI is used in reviewing your application, please contact [Name, Title] at [email address]. If you would prefer your application to be reviewed without AI assistance, please let us know when you apply and we will make reasonable arrangements.
This notice applies to all applications submitted for roles at [Company Name] [from [date] / effective [date]].
Template 2 - Interview AI
About AI use in your interview with us
We want to be upfront with you: we use an AI tool to transcribe and summarise our interview calls and video meetings. This notice tells you exactly what that means for your conversation with us.
What the AI does: With your consent, [Tool name] will record and transcribe our conversation. It then generates a summary of key points discussed (skills, experience, candidate questions, agreed next steps) which is stored in our recruitment system.
Which decisions it influences: The summary may be reviewed by other members of our team who were not on the call. This means the AI's transcription and summary can influence decisions about progressing your application.
What information it processes: Audio or video of your interview, and any personal information you share during the conversation.
What it does not do: The AI does not score or rank you based on voice characteristics, facial expressions or sentiment analysis. [Remove this line if your tool does conduct sentiment or emotion analysis — and consider carefully whether you should be using it.] A qualified recruiter reviews the summary and makes all decisions about next steps.
Your consent and your options: We will ask for your verbal consent at the start of the call before recording begins. If you prefer not to be recorded, please let us know — we will take notes manually and there will be no disadvantage to you for declining. You can withdraw consent at any time during the call by asking us to stop the recording.
If you have questions about this, contact [Name] at [email] before your interview.
Template 3 - Search & Match AI
How we find opportunities for candidates in our database
If you are registered with [Company Name] as a candidate, this notice explains how we use AI to match you to roles that may be relevant to you.
What the AI does: We use an AI-powered matching tool within our recruitment platform to identify candidates whose skills, experience and preferences align with roles we are working on. When a new vacancy is added to our system, the tool may automatically surface your profile as a potential match and alert a member of our team.
Which decisions it influences: The matching tool influences which candidates our recruiters consider for a given role. Being surfaced by the AI does not guarantee you will be approached for a role, and not being surfaced does not mean you are not suitable — our recruiters may also conduct manual searches of our database. All decisions about whether to approach you or put you forward to a client are made by a qualified consultant.
What information it processes: Your CV, skills profile, work history, location preferences, salary expectations, and any other information held in your candidate record with us. [Add: "including notes from previous conversations with our team" if applicable.]
What it does not use: The tool does not use your name, photograph, age, or location as factors in matching — only skills and role-relevant experience. [Amend this section to accurately reflect your tool's matching criteria. If you are not certain, ask your vendor.]
Keeping your record up to date: The quality of AI matching depends on the accuracy of the information we hold about you. If your circumstances, skills or preferences change, please let us know so we can update your record at [email / phone].
Your options: If you would prefer not to be matched to roles by AI — or would like to know which roles you have been considered for — please contact [Name] at [email]. Candidates in the EU and UK also have rights under data protection law to understand automated processing of their personal data; please contact us if you would like to exercise these rights.
Template 4 - Annual AI Update
For employees / existing workers
Annual notice: how we use AI in employment decisions
As part of our commitment to transparency, this is our annual update on how artificial intelligence/AI tools are used in decisions that may affect you as an employee of [Company Name].
AI tools we currently use in employment decisions:
[List each tool clearly. Example format below — adapt to your actual stack:]
1. [Tool name] Purpose: [e.g. Generating summaries of performance notes and client feedback to support appraisal discussions] Decisions influenced: [e.g. The summaries are used to inform, but do not determine decisions about performance reviews and role progression] Data processed: [e.g. Notes from your HR record, client feedback, internal performance records]
2. [Tool name] Purpose: [description] Decisions influenced: [description] Data processed: [description]
What AI does not determine: None of the tools listed above make final decisions about your employment, compensation, promotion or discipline. Those decisions are always made by a qualified manager or HR professional.
Your rights: You have the right to ask questions about any AI tool that may affect your employment. You also have the right to request that a specific decision involving you be reviewed by a human without AI input. To exercise either right, or if you have any concerns, contact [HR contact name] at [email].
When this notice will be updated: We will reissue this notice annually, and within 30 days if we adopt a new AI tool or make a substantial change to an existing one.
If you have questions about this notice, please contact [HR contact / named person]. This notice does not constitute legal advice about your individual rights, which may vary by location.
What Kyloe brings to this...
AI compliance isn't just a legal problem. It's a data problem.
The obligations in Illinois, the EU and beyond all assume something: that you know what data your AI is processing. Most Bullhorn databases contain years of legacy data - duplicates, outdated records, missing consent flags, incomplete fields. That data is feeding your AI whether you like it or not. Kyloe's suite of Bullhorn-native tools is directly relevant to building compliance infrastructure that actually holds up.
Your AI is only as compliant as your data
Deduplicate records, identify candidates with missing or outdated consent, clean legacy data before it creates AI compliance risk.
An audit-ready Bullhorn database isn't just good practice, it's the foundation of defensible AI use.
Automate your compliance documentation
Candidate notices, AI disclosure statements, consent documents — all generated and delivered automatically through your Bullhorn workflow.
AwesomeDocs is how you do it
at scale without the manual overhead.
Build your governance framework properly
Not sure where to start? We can audit your current AI footprint, identify your compliance gaps, help you write your Glass Box statement, and build an ongoing governance process into your Bullhorn workflows.
Practical, Bullhorn-specific help.
Not sure where you stand with AI?
Let's find out together.
*This guide is for informational purposes only and does not constitute legal advice. Always consult a qualified legal professional for advice specific to your circumstances. Last updated: March 2026.