ChatGPT is not your lawyer

Sunira Chaudhri

Sunira Chaudhri

Toronto Employment Lawyer

The speed of such artificial intelligence may be impressive, but its accuracy is not.

Over the last few months, I have noticed an alarming trend.


Both employer and employee clients alike have indicated ChatGPT has played a role in their respective employment negotiations and even employment agreements.


On one recent occasion, I was hired to conduct a routine employment contract review. At the outset of our Zoom call, my client quipped that the contract was likely drafted by ChatGPT.


I laughed, except, it turns out, he wasn’t joking at all.


Employers are increasingly turning to ChatGPT and other AI tools to churn out templates of employment agreements, policies and other employment related documents in an effort to save costs. This newfound dependence on ChatGPT is presumably natural given the general sentiment in the media that ChatGPT is smarter than us all and will eventually take over or, at the very least, disrupt most industries.


I have enough self awareness to know it would be a fool’s errand to encourage any of my fair readers against fully leaning into the absolute wonders of artificial intelligence. But, I will try to bend your ear on some of the pitfalls and consequences of completely relying on using ChatGPT as your free, part-time human resources manager or employment lawyer.


I created a ChatGPT account for this article and immediately was warned it “may occasionally generate incorrect information” and has “limited knowledge of world and events after 2021.” It also told me that operators of the platform could review the data I input to help improve the system, meaning I should never enter sensitive, confidential information (like employee information) into the system.


These qualifiers raised red flags for me, but I ventured on.


Employment law has evolved substantially since 2021, so from the outset I knew ChatGPT would present some serious limitations when trying to interpret this area of the law.


I asked ChatGPT, “How much should I pay an Ontario employee on termination that has worked for me for 8 years?”


The response was mixed.


While ChatGPT was able to identify the one piece of legislation that could apply to an Ontario employee, it wasn’t able to accurately identify the employee’s minimum entitlements on termination. It accurately identified the minimum notice period (eight weeks) but didn’t identify a possible minimum severance entitlement (which could have been eight weeks in this case), and perhaps of most concern, it didn’t identify what additional damages an Ontario employee could be entitled to.


If an employer followed ChatGPT’s interpretation of the law, it would find itself in serious hot water.


An employer doesn’t just need to be aware of the “minimum” entitlements it would have to pay on termination (although ChatGPT could not even accurately ascertain this amount from my basic question), but instead employers need to know the risks of total liability. Sometimes employers need to consider the reasons for termination, the appropriate timing and how other factors (for example an HR complaint or medical leave) could weigh into liability risks.


From my question, ChatGPT also didn’t consider whether the employee was federally regulated either, as is the case for all banking, transportation and telecommunications employees. Federal employees are governed by the Canada Labour Code and they are entitled to a wholly separate regime of entitlements and remedies on termination (including possibly reinstatement).


I next asked ChatGPT to draft a non-competition clause for an Ontario employee. It quickly sputtered out a lengthy restrictive covenant. ChatGPT’s speed was impressive, but the accuracy was not.


In Ontario, most employees can’t be limited by a non-competition clause at all, unless they are in the c-suite. Employers that introduce non-competition clauses for employees in 2023 contracts would be seen as out of touch but worse, would be frowned upon by the courts and even their own employees.


These are two of several examples I gleaned from playing around with the platform. It’s clear that subtle but important nuances in the law can only be caught by specialists, or experts in the employment field. Used haphazardly, ChatGPT can create more harm than good for employers and employees in workplaces.


To be fair, in most of its answers to my legal questions, ChatGPT warned me to get employment law advice. The problem is, many ignore that warning.


If you’re wondering about the employment contract I mentioned at the beginning of this column, you won’t be surprised that a slew of terms in the lengthy agreement weren’t enforceable against my client. Some of the terms didn’t apply to him at all. I proceeded to give him advice as though the contract didn’t exist because the AI-generated contract had little legal impact on his employment.


It turns out, the contract wasn’t worth the digital screen it was generated on.


Have a workplace issue? Maybe I can help! Email me at and your question may be featured in a future column.


The content of this article is general information only and is not legal advice.

More In The News

Musk vows to shrink Twitter workforce

If true, Musk will cull Twitter by 5,000 employees, leaving a skeleton staff of approximately 2,000 to run the social media company. Slashing a workforce in this cavalier manner is not only unlikely, but it would also undoubtedly cripple the operations of the company.

Read More