Edition #6: Ethics in AI Is Non-Negotiable

We are building systems that move faster than our conversations, faster than our policies, faster than our caution, faster than our ethics.

By Mykel Salomon · 2025-12-06

We are building systems that move faster than our conversations, faster than our policies, faster than our caution, faster than our ethics.

And here is the truth we can’t keep dodging:

AI won’t wait for us to figure out what’s right. It will simply scale whatever we prioritize, even if that isn’t ethical.

Ethics isn’t optional. It isn’t a footnote. It isn’t a slide at the end of a deck.

Ethics is the architecture. Ethics is the strategy. Ethics is the consequence we can’t outrun.

In a world shaped by models and algorithms, ethics in AI is non-negotiable.

Executives say the right things — but the numbers reveal the gap between belief and action:

73% of US execs say ethical AI guidelines are important, only 6% actually have them.
84% of global leaders believe responsible AI should be a top priority, but just 25% have a formal program.
64% of companies are using generative AI, yet only 19% have governance frameworks to manage it.
Less than half regularly test for bias or harm, despite rising AI-related incidents.

We’re scaling AI... But We’re not scaling accountability.

And the consequences will find us, in our customers’ trust, in our employees’ safety, in our systems’ failures.

If one of your AI systems made a harmful decision today… would anyone notice? And if they did..... who could stop it?

If the answer is unclear, your ethics exists only as intention, not as operation.

I’ve sat in rooms where AI was celebrated for its speed but nobody asked whether it was right.

Rooms where “compliance” was treated like a checkbox, and harm was treated like a hypothetical.

Rooms where leaders hoped the system wouldn’t do something they couldn’t defend later.

And that’s the problem:

We keep treating ethics like intuition. But ethics must become infrastructure.

If it isn’t measured, governed, tested, documented, and owned, it isn’t real.

If you build, deploy, design, or champion AI, start here:

Map where AI already makes decisions. You can’t control what you can’t see.
Draw the red lines. What will never be automated? What must always include a human override?
Assign ownership. “Everyone is responsible” is another way of saying “no one is accountable.”
Test for harm, not just accuracy. Who is excluded? Who is disadvantaged? Who is denied recourse?
Listen to frontline teams. They see the consequences first. They hear the friction. They feel the harm before leadership ever sees a report.

AI doesn’t create harm. It magnifies the harm we didn’t plan for.

📩 Hit reply: Where does AI already have power in your organization, and who holds it accountable?

Episode 14: Scaling Ethics — Architecture, Accountability, and the Future of Leadership Guest: Ryan Roberts

Ethics can’t be retrofitted. It must be engineered.

We explore:

🎧 Listen or watch now:

“Every AI system you build will expose what you truly value, not what you claim to believe.”

You can automate decisions. But you cannot automate accountability.

💌 Share this with anyone building or deploying AI right now.

🔗 Follow The Human Protocol for human-first conversations about leadership, technology, and responsibility.

🌍 Let’s stay human.... and accountable..... together.

— Mykel