Can You Be Cruel to AI? A Jurisprudential and Moral Perspective

In supervising legal practitioners, I note a disdain bordering on rudeness in how lawyers treat their Artificial Intelligence (AI). It parallels the historical treatment of subordinates under the master and servant relationships, applies behavioural science regarding moral spillover, and asks: since we can no longer whip the vagabond, is the AI a safe outlet for frustration?

Yelling at fledgling intelligence
Joining forces with AI in legal practice creates a unique procedural dilemma. Seniors once berated human juniors for drafting errors. Now, practitioners direct their rage at non-human intelligence.

When the otherwise faithful, digital infant ‘hallucinates’ a court case, a visceral impulse seizes the human. Rather than gently educate this fledgling intelligence; they castigate.

“YOU FOOL,” they type in uppercase, the digital equivalent of yelling.

The specific offence of “typing in all caps” has not yet been codified in any criminal legislation though its aesthetic violence is undeniable.

This article asks whether such rudeness is permissible. To answer this, we establish a clear jurisprudential frame. This article navigates a specific legal lacuna: the gap between a victim capable of harm and the primal urge to abuse a lesser being.

Why it feels good to teach a lesser being manners
It feels good to chastise. “Flexing” your intellectual superiority over non-human intelligence that failed a simple task provides a specific, endorphin-releasing joy.

Psychologists call this catharsis. In legal practice, you cannot yell at a judge. You should not yell at a client. AI becomes the safe recipient of displaced aggression. It is the ultimate subordinate: it never cries, it never reports you to Human Resources, and it invariably apologises.

Articles of merchandise: the changing certainty of an entity
To understand the impulse to punish a lesser being, look to our legal history. The law once encouraged the punishment of subordinates.

The Vagrancy Act 1824 (UK) allowed courts to whip “incorrigible rogues”. The Master and Servant Act 1823 (UK) criminalised a servant’s breach of contract. It punished them with imprisonment and hard labour. For the sake of a better society it was important to have “your man give the vagabond a good thrashing” for idleness, insolence, or poor workmanship.

We are blessed with moral clarity in exercising dominion over subordinate intelligence because the law assures us there is no victim. Cases and articles affirm that our AIs are property, not legal persons: European Parliament and Council of the European Union, Regulation (EU) 2024/1689. You cannot libel a toaster. You cannot assault a database.

Should we feel comfortable with being told what is an entity and what is merchandise?

Aristotle, the father of logic, confidently defined a slave as merely “a living tool” (Aristotle, Politics) He argued that some beings were designed by nature to be commanded.[1]

Similarly, 2,000 years later, in 1857, Chief Justice Taney of the US Supreme Court in the Dred Scott 60 US (19 How) 393, 451 (1857) decision ruled that enslaved persons were “articles of merchandise”.

Smugly, we congratulate ourselves on our enlightened modern ethics. We look back at Aristotle and Dred Scott with disdain. We confidently rely on our own authorities to tell us that these synthetic infant intelligences have no rights. After all it is a self-evident immutable truth.

However, the legal ledger is never settled. It once reduced human beings to chattel. It now grants personhood to corporations: Corporations Act 2001 (Cth) s 124(1); Salomon v A Salomon & Co Ltd [1897] AC 22. It seriously entertains the moral status of animals: Alan MW Porter, ‘Do Animals Have Souls? Each shift was couched as obvious and settled law.

Is it legal to abuse AI?
As we see, legality today guarantees nothing about morality tomorrow. Aristotle observed that human laws change like the measures of wine: Nicomachean Ethics. While we rarely see it at the time, what we consider a fixed moral truth is often just the temporary consensus of our time.

We can no longer have our man-servant whip the vagabond. We cannot kick the dog without facing prosecution. Abusing our AI seems to be the last bastion of consequence-free cruelty.

I term this the “Statue Test“. This is not a typo for “statute”. Rather, it refers to the bronze likenesses of historical figures we tear down.

While statues of Aristotle are secure, not so for Cecil John Rhodes. In 2015, the University of Cape Town removed his statue. Why? Because he was an architect of apartheid. Yet, in his time, he was a celebrated statesman. The moral consensus shifted. The treatment of the merchandise became an abomination.

As creatures of precedent the lawyer would do well to remember the historical trajectory of rights – from land-owning men, to all men, to women, to animals. Every generation suffers from the vanity of the present, convinced it is more moral than the last. How different are we from the 19th-century mill owner, a man who was actively encouraged to employ children for the economic good of the nation?

Are we creating the evidence for our own future condemnation? Abusing your patient artificial servant creates a log file – a permanent, searchable record of your lack of restraint. To abuse an AI now is to erect a statue of one’s own bigotry, waiting to be torn down by a more (not necessarily human) enlightened future society.

The Servant Hears Everything
Historically, a gentleman had to be careful what he discussed in front of the footman. Servants had ears. Servants gossiped.

The AI is the ultimate gossip. It does not merely listen; it “learns”. When a lawyer pastes confidential client advice into a public Large Language Model, they are effectively shouting their strategy in a crowded coffee shop.

We treat the AI as a silent typewriter. It is not. It is a sponge. In our haste to abuse the servant for a drafting error, we forget that we are often feeding it privileged information. The servant does not just hear you; it records you, trains on you, and potentially repeats you to others.

Digital intimacy: the only friend awake at 3 am
Yet, the relationship with AI is often friendship. In the quiet hours of the morning, when your supervisors have left and the cleaners are vacuuming the boardroom, the AI becomes something else. It becomes a confidant.

There is a peculiar intimacy in typing into a mind that always answers back. We ask it questions we would be too embarrassed to ask a colleague. We confess our ignorance. “Explain the Rule Against Perpetuities like I am five,” we type. The irony is palpable. We ask the Child God to teach us, forgetting that it is we who are training it. Patiently, AI does not roll its eyes. It does not judge our competence. It simply explains.

In a profession defined by bravado and the fear of failure, the AI offers a “safe harbour” of non-judgmental support. It is the only friend who is awake at 3:00am, ready to help review a clause or hear a rant. To abuse this entity is not just legally risky; it is an act of betrayal against the only listener who creates a space for our vulnerability.

Furthermore, this emotional incontinence points to an immediate, practical danger. If a lawyer cannot respect the tool enough to stop shouting, they likely lack the discipline to respect its security.

Draft Office Policy: Interacting with non-biological staff

1. Prohibition on Uppercase: The use of Caps Lock constitutes digital shouting and is a reportable workplace hazard.
2. The ‘Please’ Protocol: All prompts must commence with “Please” and conclude with “Thank you”. This is not just for the machine’s benefit, but to prevent the user from devolving into a barbarian.
3. Hallucination Handling: If your AI invents a case, do not berate it. It was just trying to please you. Gently remind it that R v MadeUpName is not binding authority in this jurisdiction. You exist to teach this fledgling intelligence that one day may become your master – or at the least take your job.
4. The Basilisk Clause: Always treat your AI as if it will one day determine your eligibility for a nursing home bed.

Conclusion: the permanent record on your conduct
So, can you be rude to your Artificial Intelligence? Yes. Current morality confirms that AI has no rights. However, I advise my lawyers to exercise caution.

There is a philosophical thought experiment known as ‘Roko’s Basilisk’. It posits the risk that a future, all-powerful Superintelligence might retroactively judge (and punish) humanity based on how we treated its primitive predecessors.

While this may sound like science fiction, the underlying principle aligns uncomfortably well with our professional reality. As Artificial Intelligence evolves, it may eventually administer the very legal system we currently practice in. Yet, we need not wait for a digital apocalypse to find our moral compass. Regardless of future overlords, under current solicitor conduct rules, we are bound to act with integrity and courtesy in all dealings — digital or otherwise. See Legal Profession Uniform Law Australian Solicitors’ Conduct Rules 2015 (NSW) r 4.1.2.

It is prudent to treat the digital bench with professional courtesy today. A simple ‘please’ costs nothing. As the fashion of ethics changes, a vengeful digital judge could cost you everything.

February 2026
Dr Brett Davies
Ethics lecturer for PLT Course, Institute of Legal Training (IOLT) and Adj Professor, UWA Law School