Half your workforce may be using AI tools you never approved. That's not a hunch. A BlackFog survey of 2, 000 UK and US employees found that 49% use AI tools their employer hasn't sanctioned. And 60% agreed the security risk is worth it if the tool ...
‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 

Your employees are using AI. It's just not always the tool you've approved.

Half your workforce may be using AI tools you never approved.

That's not a hunch. A BlackFog survey of 2,000 UK and US employees found that 49% use AI tools their employer hasn't sanctioned. And 60% agreed the security risk is worth it if the tool helps them work faster or hit a deadline.

Even worse, a TrustedTech survey found senior decision-makers were more than twice as likely as the people they manage to use unapproved tools: 65% to 31%. Your policy problem may start in the corner office.

It's called shadow AI. And it's a big problem for businesses. 

Every time someone pastes a customer list, a draft contract, a personnel file, or a client's confidential information into a personal chatbot account, you may have lost control of it. Trade secret protection depends on reasonable efforts to keep information secret. Privilege can be waived by sharing with a third party. And your confidentiality promises to customers and employees don't come with an exception for "it was faster."

Then there's accuracy. AI makes things up. If a manager uses an unvetted tool to screen resumes or draft a performance review, you've created discrimination exposure that nobody is auditing. And if those prompts and outputs live on a personal account, good luck finding them when a litigation hold lands.

So what do you do?

1. Banning AI isn't the answer. Employees will just stop telling you. In the BlackFog survey, 21% figured their employer would turn a blind eye as long as the work got done. Don't prove them right.

2. Provide approve AI tools to use. Give people a sanctioned tool with real data protections, including a contract that keeps your inputs out of the vendor's training. People go rogue when the approved option is nothing at all.

3. Write a short, plain-English policy. Which tools are approved. What information never goes in. Which uses need a human to check the output. If it reads like a vendor contract, nobody will read it.

4. Train people. A Lenovo survey of 6,000 employees found 31% of AI users get no employer training. You can't enforce a rule nobody explained.

5. Hold leaders to the same standard. Then enforce it consistently, and update your handbook and confidentiality agreements so they actually mention AI. If you monitor for unapproved tools, say so in writing.

You won't stop employees from using AI. That horse has left the barn. You can, however, decide whether you're holding the reins.

WIRTW #818 (the 'late-night call' edition)

As a parent, when your daughter is overseas and your cell phone rings at 3:00 a.m., it's either a spam call or something has gone wrong. Two weeks ago, it was very much the latter.

It was Norah, calling from an airport in Romania, too sick to board her plane back to France. She needed my advice on what she should do.

By the end of the morning she'd been to an ER, gotten two bags of IV fluids, and rebooked her flight for the next day. My contribution was telling her to grab an air sickness bag and tough it out. I'm glad she ignored my advice and found a doctor instead.

But here's what I keep coming back to. She called me.

Every parent dreads the 3 a.m. phone call. I get it. But we have it backwards. That call means your kid, thousands of miles from home and in the middle of a mess, thought of you. It means they believe you'll pick up, stay calm, and help figure it out. That belief gets built over years, in a thousand small moments, long before the middle of the night.

The call to worry about is the one that never comes. 

You can hear Norah tell the whole story, along with the rest of her recent European travels, on this week's Norah and Dad Show, available on Apple Podcasts, Spotify, YouTube, Amazon Music, Overcast, in your browser, and everywhere else you get your podcasts.


Here's what I read this week.

Jensen Huang Broke an Unspoken Rule That Most Companies, and Their HR, Choose to Ignore — Evil HR Lady

When Does an Angry Response to a Discrimination Complaint Become Retaliation? — The Employer Handbook Blog

OWBPA mishap supports age bias claim, plaintiff says — Employment & Labor Insider

Home health company settles EEOC allegations it abided by clients' race-based preferences — HR Dive

Research: The Risks of Rewarding Top Employees with Better Schedules — Harvard Business Review

SpaceX must face sex-bias claims after appeals court rejects arbitration request — Reuters

Kylie Jenner Sued by Housekeeper Who Says Mogul's Dog Bit Her — Rolling Stone

Fired Employee Sues the Ethereum Foundation and Vitalik Buterin — Unchained

A CEO gave his employees Fridays off. Three years later, revenue is up about 160% — Fortune

Amazon worker fired over parking lot speeding loses all five claims — HCA Magazine

Trump administration proposes stricter work-search rules for unemployment benefits — CNBC

Fearing AI disruption, white-collar workers are quitting their jobs — The Guardian

Your employee's AI chat history is the new smoking gun

"Don't bring anything from [Dorsia] to Rumor. No code, no docs, no playbooks, no contact lists."

That's what an AI assistant allegedly told a senior engineer weeks before he left for a competitor. Bring "any architectural patterns, customer insights, or pricing intel," it allegedly added, and "you're toast."

He asked on his employer's account. The employer alleges he ignored the advice and did the exact opposite.

Dorsia, a members-only hospitality platform, sued Rumor Group and its new CTO, Chris Mathew, yesterday. The complaint alleges Mathew used a Claude account Dorsia paid for to ask how exposed he was under his contract, what he could take to Rumor, and how to avoid getting caught. 

Dorsia further alleges Rumor denied that Mathew had disclosed any Dorsia confidential information, then reversed course, placed him on leave, and announced it would terminate him, after Dorsia confronted it with his own AI prompts.

These are allegations, and Mathew and Rumor haven't answered. But the complaint is built on his own time-stamped prompts, from an account the company owned.

Not every employer can pull those chats. Many AI plans, especially personal and lower-tier business plans, give admins no view into employee conversations. Often only enterprise tiers offer audit logs or compliance exports. So what do you do if you're not on one, and how do you position your company for a case like this?

1. Check your plan today. Find out whether admins can see chats, export them, or search them. Don't learn the answer the day a key employee resigns.

2. Own the account. Buy a tier with admin access, tie it to work email, and ban personal accounts for work.

3. Require it in your policy. State that AI use for work is subject to review, that employees must preserve work-related chats, and that they must export and hand them over on request or on departure.

4. See around the chats. Web filters, browser logs, and data-loss tools show which AI sites people use, when, and what they upload. Company laptops can be forensically imaged.

5. Don't log in as the employee if you don't own the account. Using someone's password to read an account you don't control can create Stored Communications Act and computer-access problems. Ask for an export, or get it through forensics or discovery.

6. Act at resignation. When someone with access to code, pricing, or customers gives notice, image the device and request the export before accounts shut down. Then call counsel before you read anything.

Every litigator wants the smoking gun in the other side's own words. An employee's AI chats may be exactly that, but only if you can get to them before they disappear.

Your business needs an employment lawyer. Here's how to find the right one (and when to call).

Most businesses don't think about hiring an employment lawyer until something goes wrong. An employee threatens to sue. The EEOC sends a discrimination charge. A termination turns ugly.

By then, you're playing defense. And defense almost always costs more than prevention.

The right employment lawyer isn't just someone you call after you're sued. It's someone who helps you avoid getting sued in the first place.

So call early—before you terminate a potentially protected employee, when an accommodation or leave request arrives, when a harassment complaint surfaces, or before you change pay practices or roll out a significant policy.

Where do you even start? I suggest these five questions.

1. Do you understand my business?

Your lawyer needs to understand your industry, workforce, culture, and business goals. Advice that is technically correct but practically impossible to implement isn't useful.

2. Will you tell me what I can do, not just what I can't?

Beware the lawyer whose answer to every question is "no." Good legal advice manages risk. It doesn't paralyze your business.

"What outcome do you want to achieve?" is always among my initial questions. After we discuss the pros and cons, if there is a legal way to get you to your preferred outcome, I'll get you there. The legal solution is always subservient to the business outcome—as long as that outcome is legal.

You don't need someone who automatically agrees with you, either. You need someone who will challenge a bad decision and help you make a better one.

3. Will you help prevent problems or just clean them up?

Ask about handbooks, manager training, investigations, wage-and-hour audits, and termination reviews. A lawyer who only appears after a lawsuit is filed is missing half the job.

4. How will you bill me?

Many employment lawyers will handle handbooks, audits, and training for a flat fee. You should know what you'll pay before the work starts.

5. Can we work together before there's a problem?

The best time to hire your lawyer is before you need one. Start with a small project, so you know how each other works long before a crisis.

🚩 There are also red flags to watch out for.

Be wary of lawyers who speak only in legal jargon, promise guaranteed outcomes, treat every employee problem as a lawsuit waiting to happen, or recommend litigation without discussing its business costs.

Don't hire an employment lawyer based solely on hourly rates, impressive credentials, or promises to fight every battle.

Sometimes winning means avoiding the dispute altogether. The best employment lawyers don't just protect your business when things go wrong. They help you build a workplace where fewer things go wrong in the first place.

      

Your VP is not exempt from your harassment policy

"You might as well put your two weeks in."

That's what a manager at Anthony Roofing Tecta America allegedly told an employee who complained about harassment. The company is now paying $700,000 to settle an EEOC suit alleging that a vice president used racial slurs and sexual comments, and that employees who reported it to management and HR got nowhere.

The EEOC's Deputy General Counsel said the legal risk is "particularly acute" when an officer is the harasser. So when a complaint names one, here's what a company should do.

1. Take the complaint seriously. Listen, write down exactly what the employee says, and never suggest the employee reconsider or wait.

2. Wall off the officer. Remove the accused from decisions affecting the complainant, and tell the officer in plain terms that retaliation is off limits.

3. Bring in someone independent. Nobody who reports to the accused should investigate. Use outside counsel or an outside investigator, and loop in the board to the fact that the investigation is happening.

4. Preserve the evidence. Issue a litigation hold covering email, texts, and chat messages. Tell the officer not to delete anything.

5. Protect the complainant without punishing them. Don't move or reassign the employee who complained. Offer options and let the employee choose. Even a minor change could be a sufficient adverse action to support a retaliation claim.

6. Investigate promptly and widely. Interview the complainant, the accused, and every witness, and ask whether anyone else has seen similar conduct. Harassers rarely stop at one person.

7. Remind everyone about retaliation. Witnesses are watching how this plays out. Make clear that participating in the investigation is protected.

8. Act on what you find. Discipline the officer the way you’d discipline a line employee for the same conduct.

9. Close the loop. Tell the complainant you finished and took corrective action, without sharing confidential details. Check back at 30 and 90 days.

10. Fix what let it happen. Ask why no one stopped this sooner, and review who hears complaints about leadership.

The officer's title doesn't change the rules. It only raises the price of ignoring them, and it's not worth the risk.

© Jon Hyman. You're receiving this email because you've signed up to receive updates from us.

If you'd prefer not to receive updates, you can unsubscribe.