By Chuck Gallagher — Business Ethics Keynote Speaker and Trainer
TL;DR: Chuck Gallagher, AI ethics speaker and author, examines two 2026 lawsuits against Meta and Workday to argue that no company can hand its moral accountability to an algorithm — the person who runs the tool still owns the outcome.
When the Machine Does the Work, Who Owns the Outcome?
Somewhere this week, an employee sat down at a screen and signed a document she did not fully read. It was an addendum to a policy that had been on the books for years — the acceptable-use-of-technology agreement every new hire signs and forgets. Only this time there was a new page. It said, in plain terms, that the artificial intelligence tool on her desktop does not understand meaning, values, or context the way a person does. It said the tool predicts patterns, not truth. And it said something she may not have caught. She is responsible for every input she gives that tool, and accountable for every output it hands back.
Read that last part again. The machine does ninety-nine percent of the work. She owns one hundred percent of the consequence.
That is the quiet bargain being written into workplaces across America in mid-2026, and most people signing it have not thought it through. As an AI ethics speaker and author, I have watched this shift up close, and here is what I keep telling boardrooms: the tool got smarter, but accountability did not move. It stayed right where it always was. With you.
Now, two lawsuits filed this summer show what happens when nobody can point to the human who made the call.
On July 14, 2026, twenty-six current and former Meta employees filed suit in federal court in Oakland, California. They allege that Meta used a web of internal AI systems — performance rankings, productivity scores, even dashboards tracking “AI token consumption” — to decide who got cut in a layoff of roughly eight thousand people. The trouble, they say, is that those metrics cannot be earned by someone out on medical or family leave. You cannot rack up token usage from a hospital bed. The plaintiffs claim workers on protected leave were flagged as low value and shown the door. Meta denies it flatly, telling CNBC the claims lack merit and that workforce decisions were and are made by people, not AI. On July 17, a federal judge declined to block the layoffs, but left the door open to revisit if evidence shows the algorithm drove the decisions.
Let me be clear about what that case is and is not. Nobody has proven Meta did anything wrong. These are allegations. But notice the shape of the fight. Meta’s whole defense rests on one sentence: people made these decisions, not machines. That is the sentence every company is going to have to be able to say — and mean.
The second case cuts even closer. On June 22, 2026, a federal judge in San Francisco ruled that Workday must face claims its AI hiring software screened out applicants in ways that allegedly violated California and federal anti-discrimination law. The suit, brought by Derek Mobley, has been winding through the courts since 2023. Here is the part worth sitting with. The judge did not rule that Workday’s software discriminated against anyone. She ruled only that the case can go forward. But in letting it proceed, the court treated the vendor behind the algorithm as a possible participant in the hiring decision — not just a neutral tool the customer controls. That is new. And every company buying screening software should feel that one in the gut.
So what do you actually do about it? Because pointing at the problem is easy. Living inside it is not.
The good news is the answer is not a mystery. Recent research offers plain checklists any organization can adopt. One is the CLEAR framework for prompting: be Concise, Logical, Explicit, Adaptive, and Reflective. The reflective part is the whole ballgame. It means a mandatory human-in-the-loop review of anything the machine produces. Never upload proprietary data or personal information. Treat every AI output as a rough draft, not a verdict. Another is the C2V2 governance model for teams handling data: Control your sources and consent, test for Consistency and bias, keep the work centered on human Value, and guard the Veracity of what feeds the model. Audit the inputs. Question the outputs. Keep a person in the chair.
I have said this in a hundred rooms, and I will say it here. As an AI ethics speaker and author, the line I come back to is the one I built a whole life’s work around: every choice has a consequence. The choice to let an algorithm rank your people is a choice. The choice to skip the human review is a choice. The choice to sign the addendum without reading it is a choice. And the consequence does not care how good the technology is.
Here is the hard truth underneath all of it. AI did not create a new ethical problem. It just made the old one faster and harder to trace. Doing right by people — the ones you hire, the ones you fire, the ones your software quietly sorts — is the oldest question there is. The machine can crunch the data. It cannot carry the weight. That part is still yours.
Trust me. Someday a lawyer, a reporter, or your own conscience is going to ask who made the call. “The AI did it” will not be an answer. It never was.
Frequently Asked Questions
Can a company blame AI for a bad hiring or firing decision?
Not really. Current legal and ethical thinking holds that a human stays accountable for what an AI tool produces, even when the tool does most of the work. The Meta and Workday cases in 2026 both turn on whether a company can shift responsibility to an algorithm, and so far courts are skeptical.
What is the CLEAR framework for AI prompting?
CLEAR stands for Concise, Logical, Explicit, Adaptive, and Reflective. It is a checklist for writing safe, effective prompts, with the reflective step requiring a person to review every AI output before it is used. The goal is to keep a human in the decision, not hand it to the machine.
Did the Workday ruling find that its AI discriminated against applicants?
No. On June 22, 2026, a federal judge ruled only that the case could move forward, not that Workday’s software actually discriminated. The significance is that the court allowed the vendor behind the algorithm to be treated as a possible participant in the hiring decision.
How can my organization use AI responsibly without getting sued?
Start with a written policy that makes employees accountable for AI outputs, require human review of anything the tool generates, and never feed proprietary or personal data into a public model. As an AI ethics speaker and author, I tell leaders the safest posture is simple: treat every AI result as a draft a person must verify. Frameworks like CLEAR and C2V2 give you a repeatable checklist.
Why do people say AI ethics is really just old ethics?
Because the core question — what do we owe the people affected by our decisions — predates AI by thousands of years. Machine learning did not invent the problem; it just runs it faster and at a larger scale. The duty to treat people fairly stays with the humans in charge.
Bring This Conversation to Your Organization
The companies that come through this era with their reputations intact are the ones that decide, now, that a human stays accountable for every automated call. That is not a technology problem. It is a leadership problem, and it starts in the room where the tone gets set. If you are building an AI policy your people will actually follow, or you want your team to feel the weight of every choice having a consequence before a lawsuit teaches it to them, that is the conversation Chuck Gallagher brings to boardrooms and stages. Start it at ChuckGallagher.com.
Five Questions for Reflection
1. In your organization, if an AI tool helped make a hiring or firing decision today, could you name the person who is accountable for it?
2. Have your employees been asked to sign responsibility for AI outputs they may not fully understand, and were they trained before they signed?
3. Where in your workflow does a human actually review what the machine produces, and where has that review quietly disappeared?
4. If a reporter or a court asked how your AI tools score people, could you explain it clearly, or would you discover the answer during discovery?
5. What decisions in your business should never be delegated to pattern-matching, no matter how good the technology gets?
