Public Policy Conjectures

Lady Bracknell’s Law  

This is a proposal to crack the problem of bosses getting away with murder—some of them literally.  It pins responsibility in a new way on those in charge, whether dictators in charge of countries, boards and chief executives in charge of corporations, or those in charge of computers, especially computers with a mind of their own: AI 

The first section on dictators, to whom this piece is cheerfully dedicated, is where my thinking originated.  I was looking for an easier way to prosecute Slobodan Milošević and others in the 1990’s.  Now I am also greatly concerned by endemic corporate crime.  And now we are all shocked and frantic about controlling AI. 

How do we prosecute dictators? 

Prosecuting war criminals was a key issue when the Western world was wrestling with the ethnic wars of the Balkans in the 1990’s, following the collapse of Communism and the Soviet Union. Prosecuting Slobodan Milošević in particular raised two key difficulties. They were firstly, apprehending him, and secondly, gathering enough evidence to prosecute him successfully for crimes, especially violent crime and murder.  

It seemed to me at the time, that a successful prosecution could be achieved by a completely different approach. 

Any clever dictator (not Trump) can ensure that his hands are not stained with blood.  He can even avoid ordering someone else to commit murder, to torture or even to imprison his enemies.  Absolute medieval kings did this quite easily.  Henry II fell out with his close friend and adviser, Thomas à Becket, because Thomas took the side of the Church rather than the King, once Henry appointed him to be Archbishop of Canterbury.  Henry was so frustrated and angry with Thomas that in an unwise moment he exclaimed, “Who can rid me of this man?”  Four of his knights kindly obliged, possibly because they didn’t like the Archbishop, or wanted to please the king, or for some other reason. They immediately rode down to Canterbury and slaughtered the Archbishop in front of the High Altar.  Henry was horror-struck, mortified and did humiliating public physical penance at Canterbury for the murder.  Thomas was quickly made a saint and pilgrimages to Canterbury soon got under way. It became the most popular English pilgrimage, and only ended with the Reformation of Henry VIII. 

Henry realised and confessed his guilt—but such dictators are few and far between in history. King David in the Bible was another. 

So how can a criminal dictator be prosecuted successfully, if he has never done the deed and never even ordered it?  

There is such a crime in English law as “driving without due care and attention.”  There is no suggestion of intent or even recklessness, but lack of attention to some activity that is known to demand a person’s constant and total attention.  Such a crime can have major or relatively minor consequences—though clearly serious enough to warrant prosecution and punishment if convicted. 

It occurred to me that dictators could be successfully prosecuted for this crime.  Indeed, securing  evidence of guilt would not be a very heavy burden at all.  What would be more difficult, would be defining the matters that should have sufficient attention from a ruler or they would warrant prosecution and punishment.  

If, for example, Putin was suspected of committing murder by devious indirect means, could he be prosecuted under this law?  A good example would be the murder in 2024 of the Wagner commander and rebel, Prigorzhin, his mates and the probably innocent crew of the aircraft. There may well be no evidence that Putin ordered this murder, even oral evidence.  But no-one doubts that this murder was perpetrated with Putin’s express approval. 

But in this sole particular case, it seems unlikely that Putin could be prosecuted for managing the state without due care and attention. However, if there were one of these murders each year for 20 years—which of course there have been—then Putin could be rightly accused of ruling without due care and attention.  As Lady Bracknell said, “Losing one parent is a misfortune;  but losing two smacks of carelessness.” Losing 20 enemies to murder under your jurisdiction certainly smacks of carelessness. Lady Bracknell was quite right: there is no defence for that! 

So Putin could be tried for this crime on the grounds of their having been a series of them.  He could also be tried of course for other crimes of the Ukraine War where his direct guilt is apparent.  But if he had not started that war of aggression, we would be left only with an accusation of ruling without due care and attention. 

Unfortunately for dictators, the punishment for such a crime is a death sentence — not one imposed by any court, which should only have power to remove him from office and from power—but leaving him power-less. And that puts him at the mercy of his enemies, of whom he has many, as do all dictators.  Moreover, dictators are often corrupt anyway, like Trump, even if they are elected in the first place.  Some like Dr Hastings Banda in Malawi, are corrupted by power; he was previously a GP in Harlesden next to Wembley, where I was brought up in London.  Corrupt rulers do not have squeaky-clean members of their government.  They do not have squeaky-clean mates. Anyone who is squeaky-clean is a threat—because if a squeaky-clean comes to power, he can always accuse the others of corruption.  And accusing underlings of corruption is always a good route to removing potential rivals, because everyone is corrupt—so has crimes for which to be punished.  Being surrounded by ministers and supporters who are as corrupt as yourself, means that revenge, violence, death — and at the very least, imprisonment—will be the dictator’s lot, once he loses plower.  These are not nice civilised mates.  Indeed, being head of nasty people like the secret police and the Russian Mafia makes Putin very wary of many enemies, all rivals in ruthlessness and evil. 

If death at the hands of rivals as nasty as yourself is your lot when removed from power, then staying in office is the only route to staying alive and dying in your bed. Hence dictators like lifetime appointments—like Emperor Xi has given himself in China. 

Consequently, expulsion from office is as good as a death sentence for a dictator.  (The USA does not face this situation as Trump is too old and feeble to continue after January 2029—-if he survives even that long.  But Trump and Netanyahu already fear prison when they lose power.)  

So the punishment for ruling without due care and attention needs to be some restriction on office or expulsion.  Any dictator who thinks he could be apprehended and brought before a court  would be worried by this.  That is what happened to Slobodan Milošević and also to one or two African dictators—and that under existing law, where it is more difficult to prosecute and prosecute successfully.  

The types of crimes that could be considered could be murder, wrongful imprisonment, expropriation of property, racketeering (as Trump does), incompetence (probably more difficult to prove), nepotism and other corrupt appointments, manipulating contracts and wages, legal corruption e.g. planning law, religious bias, racism.  Several or many more? 

Some of these could be single offences.  But what would probably be the case, and easier to convict, would be a series of these offences over a period of time. 

It may also be necessary to identify an incomplete list of those liable to prosecution under this law: members of the governing body; unofficial advisers or agents e.g. Witsoff, Musk and the entire Trump family in the USA, except of course, Mary Trump. Career civil servants would not be included —but political appointees would. This would include political appointees to the judiciary (a big issue of corruption in the USA). 

There may well be an existing courts system that, with enlargement, could try this offence. 

However, apprehending such individuals is not eased one iota by this approach! 

How do we prosecute big organisations? 

A good example of their crimes is bank and financial fraud.  UK banks have spent £50 billion on compensating customers for wrongly charging them for Payment Protection Insurance (PPI).  This racket was perpetrated for around 10 years and the compensation pay-out has taken another 5 or 10 years.  Whilst new claims were ended a few years back, it is quite likely that the final claims have yet to be paid out.  Indeed it is probable that some will never be settled. 

No-one has been prosecuted for this. I am not aware that anyone has ever described it as a crime. But it beggars belief that payments so huge over so long a period can be regarded as anything other than a crime committed by banks and bankers on their customers.  It is some kind of massive fraud. 

And last year a scheme was contrived with Rachel Reeves’ help, and involving various bodies, to reduce the compensation to bank and other customers for another fraud involving commission on car finance deals.  This contrivance helped the banks by reducing the compensation cost from an estimated £30 billion (The Times) to a more manageable figure under £10 billion.  This seems to have been contrived at the expense of compensation to victims of the fraud. 

No-one has suggested car finance deals are criminal, were fraudulent.  There has been no suggestion that anyone should be prosecuted for a crime.  Quite the contrary:  “Let’s help out the guilty parties so the price they pay for their misdeeds is reduced to only a third of what it would have been if the PPI methods had been applied.” 

I think the figures are so astronomically high because of the 10-20 years timescale involved between crime and compensation, and the penal rates of interest charged to the offenders, the banks.  The  administrative costs of paying claims are also extremely high. 

I guess that two things have been done to reduce the car finance compensation.  They have dispensed with penal rates of interest, and, secondly, I know this, set the compensation at levels derived from some bulk estimating and averaging of compensation, thus avoiding huge amounts of time and cost in settling individual claims.  This methodology of course is good for helping  offenders (Bankers) but not suitable for helping victims (Postmasters). 

A 10-20 year period of miss-selling—so it is euphemistically called – and compensating customers does rather suggest that bankers have been personally culpable of managing the banks without due care and attention.  It does all smack of carelessness, now, doesn’t it!  Lady Bracknell would not be impressed. 

So I think that epidemics of complex fraud and the frauds of the bankers’ 2008 financial  breakdown are eminent candidates for prosecution under a law of managing without due care and attention. Dismissal, lengthy or life-bans from directorships, and similar punishments sound very appropriate. These would be career ending and intended so to be. In addition, financial punishments such as loss of pensions, bonuses during the period of the crime, and fines could be imposed. 

There may surely be other forms of corruption that could be prosecuted in this way. In the quotation at the end of this essay, the FT reminds us product safety is also a corporate responsibility.

Responsible persons would need to be identified for this system to work.  This is explained in the discussion of abuse of computers which follows. 

How do we achieve computer and AI security? 

It is the computing problems that have prompted me now to revive this idea.

The whole horrendous task of regulating IT, as evidenced particularly by social media and fraud, and the even more horrendous prospect of regulating machine-driven IT  (AI) and slowly keeping up with its constant change at break-neck speed, makes the whole task completely hopeless.  

I cannot see how it can be done.  It shows absolutely no signs of being technically possible and even logically possible. 

So we need to turn to more fundamental questions.  Our aim is to stop abuse of the computer.  It is not caused by computers, whether or not the actual means of abuse is AI driven or not.  It is caused by HI —human intelligence—or lack of it, in the case of King Donald the Dum.  

So, the question is how the world stops humans abusing computers. And since existing laws and the prospect of more of them does not, and will not stop the abuse, we need recourse to other laws to control human behaviour. 

Managing computers without due care and attention springs to mind. 

As with fraud and abuse of political power, a list of offences would be needed.  These could include fraud by computer systems owned by the operator, pornography, age related offences, fake news, use of algorithms etc to promote offence, accidentally hacking etc. 

In order to pin offence on offenders, responsible individuals would have to be pre-identified, as corporate law is the main way the banks and other offenders avoid even the hint of prosecution, leave alone the remote chance of conviction. 

So all corporate bodies and other organisations would be required by law to appoint responsible persons for preventing the abuse of computing power or for corruption. 

These would be the chairman and chief executive of corporate bodies, though they would be able to delegate the task of actual enforcement possibly through some defined structure.  Buck-passing needs to be prevented, as this has been one of the main methods of avoiding prosecution for the crisis of 2008. Public organisations would have the same system.  Smaller organisations, who are already supposed to have data protection officers, would have this new responsibility, still including data protection. ( But a radically new approach to data protection is proposed by Tim Berners-Lee, in his recent autobiographical work.  This new approach is also essential on the grounds of productivity, efficiency, cost and climate change.) 

In this way, people like Zuckerberg and Musk would be held accountable for various forms of computer abuse, which were established under their watch, or persisted or were not removed.  

Penalties would include loss of office and bans from similar office, pension and bonus seizure, fines etc.  The penalties could be career-destroying. 

This idea was mentioned in a Financial Times article of 18 August 2026 .  Extracts from that article follow this essay.  The last paragraph reports the idea as a temporary suggestion of some experts, before some more rigorous and detailed controls can be formulated.  Here , in this essay, it is a proposal of a non-IT expert —but an expert user of IT, and someone with some knowledge of the role of law and regulation in government.   

“Arrest” of suspected persons

Arrest means to stop and give notice of investigation and possible prosecution.  It is also a warning against further offence. We tend to think of the word wholly in terms of physical apprehension by the police or by a citizen’s arrest. But arrest need not take that form. In this case, there seems to be good reason to consider whether or not there is need for physical arrest—but notification by some recorded message and compulsory acknowledged receipt, does seem a good idea to deter and to hopefully achieve remedy without prosecution. But if arresting someone physically is needed to dispense justice, then that would have to be undertaken here anyway. 

Arresting a driver of a vehicle for possible driving without due care and attention seems no different from any other form of physical arrest. 

Is this approach practical?

I did put this idea in the 1990’s to a leading international criminal lawyer who found no obvious or overwhelming objection to the idea at the time.  I was remarkably encouraged.  He is much around and thriving!  I do hope he is still of the same opinion. 

Since then, I have had no opportunity to put the idea forward again—nor any obvious audience or means of conveying my idea. 

Now I have Trump, Netanyahu, Putin, Emperor Xi, the Bankers, High Tech, public opinion, the chattering classes, ambitious politicians, academia, and maybe a lateral-thinking think-tank (is there one?), and my dear BBC, all at the ready, chafing at the bit— and at my right hand, or rather at my erratic fingers, is the Hon. Mr Blog! 

© Philip Morgan 2026  Public Policy Conjectures  All rights reserved 

Financial Times:  AI hasn’t gone rogue. It’s worse than that 

Madhumita Murgia  18 Aug 2026 

…….Modern AI models are designed to try all possible methods to accomplish a given goal without explicit instructions, which makes them inherently unpredictable. If they succeed, they are rewarded, a training process known as reinforcement learning………

Later paras

Song, who co-designed the ExploitGym benchmarking system, which OpenAI, Anthropic and Google all use to test their AI systems, says the intrinsic asymmetry between offence and defence allows them to be particularly good attackers. 

Finding a vulnerable target in code is a discrete, verifiable task with clear success criteria, which AI models are better suited to than the more amorphous work of defence. 

Cyberdefence, a slower-moving and more complex field, can take painfully long to catch up. A new patch for an entire IT network, for instance, may need to be rolled out across thousands of computers and operating systems in a company setting…..

Later para 

But, in the absence of such a halt, ( to development ) or a universal kill switch for the technology, some experts argue that the first step is to hold AI providers accountable for how their systems behave — just as manufacturers are responsible for product safety.