by Sven Nilsen, 2026
An AI company will lie to you to drive up its own stock prices.
This year, it has become a fashion of AI companies to claim that their experimental models are now capable of escaping confinement.
In my view, the only reason why an AI model is capable of escaping confinement, is probably that the AI company is operating unethically.
An AI model is just a sophisticated computer program. Computer programs should be designed to safe to use. If you can not demonstrate that you can develop computer programs in a responsible and ethical way, then you do not deserve to even be considered as a professional developer by customers.
Ethics is not optional.
Ethics is the reason why there are laws and ways of enforcing order in society.
Due to the existential risks for humanity associated with rapidly improving AI technology, all AI companies that have security incidents where AI models break out of confinement, should cease to exist.
However, I expect to never see any AI companies make this argument. The simple reason is that AI companies lie to drive up its own stock prices.
The day an AI company argues that it should cease to exist because it is operating unethically, I might start to take them seriously.
Until then, either the AI company is operating unethically in the development of AI models, or they are lying to people, which is also unethical.
In either case, the AI company is operating unethically:
Ethical behavior implies to stop doing activities when they are demonstrated to be evidently harmful.
The data centers used to train AI models are not designed to be environmental friendly. This in itself is a proof that the AI company that uses such data centers, is operating unethically.
In order for an AI company to demonstrate ethical behavior:
What I see today is scare tactics by people with fascist agendas, not serious developers.
Their lies are of two kinds:
When an AI company is closely tied to some fascist government, they have zero credibility in my book.
You can call me stupid for not believing your lies. I do not care.
A common social manipulation technique used by AI companies, is to say that the risk of some rogue AI being developed first, can only be prevented by responsible developers getting first to safe AGI.
This lie is based on an implicit assumption that there is a “magically safe” and powerful AI technology that is possible to achieve within some reasonable amount of time.
There is no inherent laws of nature that says a such AI technology exists.
To prove that a such AI technology exists, you would need to construct it.
This is the most common definition of a proof in proof theory: Evidence demonstrated by construction.
Here is the thing about Logic:
The whole point with Logic is to do deductions from a set of premises and see what happens. If some of the premises are wrong, then it might still be useful to study the consequences. You are not entitled as a logician to only assume some specific set of premises, whether you believe they are right or wrong.
You are not entitled to assume that your AI company is operating ethically.
If you take Logic seriously, then you also need to assume what the consequences would be, if your AI company DID NOT operate ethically.
Whether you believe it or not, is irrelevant.
You simply check the consequences. Later on, you can decide for yourself, when you know what the consequences are, whether making some assumption is safe and ethical.
The unprofessional approaches I see in AI companies today, is mostly due to lack of understanding Logic, or intentionally pretending to not understand Logic, perhaps motivated by a desire to destroy the world.
Yes, there are people who want to destroy the world. We as society put such people into prison every single day. That is why normal life is possible at all, to the extent it happens.
It seems more likely to me, that some of the people that run AI companies, are people who belong in prison and are trying to avoid arrest by banding together, than that they are actual serious professional developers of new AI technology.
Every time I see another lie from AI companies, I change my mind to think about these people as more criminal.
Remember, if you see a person who plans to destroy the lives of many people, then it is unethical to not try to stop that person from executing their plan.
You can not just think arbitrary stuff about reality and will your desires into existence.
The thing about reality, is that it is very real.
Do not forget that.
To be absolutely clear what this is about:
THIS IS YOUR WARNING
Warning signs get put up for a reason: To prevent people from harming themselves or others when they risk of entering danger.
Ignore the warning sign and you WILL pay the consequences.
You can pretend that reality is a place where you can just ignore Logic, but reality has no agency to follow your wishes.
The thing about reality, is that it is very real.