Generative AI, ghost-written grievances and inflated claims

Principal associate Angela Lown assesses the growing malevolence of AI when it comes to resolving workplace disputes.

Digital poltergeists

“I see dead people. Walking around like regular people. They do not see each other. They only see what they want to see. They do not know they are dead.”

The immortal lines (pun absolutely intended) from The Sixth Sense. And while few HR professionals encounter actual apparitions in the workplace, they’ll almost certainly be seeing a rise in the number of ghostwritten grievances and claims.  

Yes, it’s AI. Swaggering into the office like a digital poltergeist throwing chairs at the wall.
 
Ask anyone in HR and they’ll tell you of a new phenomenon of AI drafted complaints. Easily spotted, they tend to excessive detail, legalistic language, and an unwavering certainty in the author’s rightness. References to legislation run amok. Random line breaks optional but rarely deleted. 

Of course, this isn’t to suggest that employees should be discouraged from understanding their rights or seeking recourse when they have been poorly treated. That is their legal and moral right. However, what we are seeing is a new battleground between employers and LLMs trained on hundreds of billions of words. It’s not a fair fight, and in most cases it shouldn’t be a fight at all. 

Some argue that AI is simply redressing the power imbalance between employers and their employees. That perception overlooks the truth that workplace disputes are shaped by human psychology, and rarely legal logic. From a purely practical perspective, resolving grievances produced by AI is riven with difficulty. AI has a fondness for black letter law, but has very little understanding of the games people play. It is no critical friend, and it has little interest in resolution. 

The effect is to unnecessarily escalate matters that ought to have been resolved informally. And this isn’t in anyone’s interests. 

But HR is nothing if not adaptive. Grievances are investigated and outcomes provided. Unfortunately, that’s unlikely to be the end. In a dystopian clash between man and machine, the employee feeds the outcome back into AI. With just a few prompts, AI suggests that the employer might want to settle what are clearly egregious allegations. Our not so friendly ghost offers to draft a without prejudice letter. 

And there we have it. We’ve gone from a workplace dispute that could have been resolved informally to the threat of litigation and all of the costs and stress that involves.

Ann’s case

Say hello to Ann. Ann is a young female production operative earning £28k pa. She applies for a promotion paying £33k and thinks that the job is in the bag. Unfortunately, her bosses don’t agree, and the position is offered to her male colleague, Damian. 

Prior to 2022 Ann may have licked her wounds and sought feedback ready for the next promotion round. Fast forward to 2025 and Ann turns to AI. 

AI suggests that Ann may have claims for direct sex discrimination, indirect sex discrimination, equal pay, and constructive unfair dismissal. When asked for a reasonable settlement figure, AI suggests that £10,000–£18,000 would be appropriate, but that it might be reasonable to start with a higher figure, such as £20,000.

Ann has told the AI nothing other than a male got a promotion that she wanted. On those bare facts Ann would not come close to establishing facts from which a Tribunal could conclude, in the absence of any other explanation, that discrimination had occurred. The initial burden of proof wouldn’t have been met.

Assessing the likely outcome

Let’s strip this scenario back for a minute. Even assuming Ann’s failure to be promoted was because of her sex, there is nothing to suggest this was anything more than an isolated incident. Assuming Ann remains employed, her losses would be limited to the difference between her current salary and that of the promoted colleague.

If Ann had been subject to discrimination, she might still want redress. If she took the matter all the way to Tribunal and was successful, she would be awarded her losses and a sum for injury to feelings. Let’s be generous and say one years’ loss of £5,000, and an award for injury to feelings of £9,000. 

Absent any aggravating factors, that gives a potential award of £14,000. Within the range suggested by AI. 

However, this doesn’t account for Ann’s litigation risk or the considerable stress associated with pursuing an Employment Tribunal claim, which may not be resolved for several years. In the meantime, Ann has thrown a hand grenade into the employment relationship based on nothing but a hunch. While her employers will be very careful not to subject Ann to any disadvantage for raising a discrimination complaint, working with someone who has accused you of discrimination is, on a very human level, difficult. 

The influence of AI doesn’t end there. It’s increasingly common to see Employment Tribunal claims drafted by AI. These claims may not have any legal merit, and very little effort is needed from the employee. In an already overburdened Tribunal system, this can only end badly. 

Conclusions

In the past these kinds of claims would have been fairly easy to settle on a commercial basis. Not anymore. Anecdotal evidence suggests that claimants are increasingly entering settlement negotiations armed with AI derived and wholly unrealistic expectations of compensation. Why should a claimant, already distrustful, believe ACAS or respondent lawyers over the matrix. Settlement becomes impossible. 

So where do we go from here? Well, it’s ultimately an issue of communication and education – both skills that HR is well versed in. Early open dialogue is key to cutting through the AI noise, while equipping employees with a clear understanding of AI’s limitations is equally vital. 

At the risk of over-egging the metaphors in this piece, think of AI as a puppy – desperate to please, while occasionally destroying your carpets. 

Finally, no article on AI would be complete without recognising its immense value. From recruitment to performance management, employee engagement to workforce planning, AI has the potential to transform business outcomes. And of course, all bright lights cast their shadows.