> Mobil and other oil companies are also shown to be advertising directly against electric cars in national publications, [...] Chevron bought patents and a controlling interest in Ovonics, the advanced battery company featured in the film, ostensibly to prevent modern NiMH batteries from being used in non-hybrid electric cars.
> car makers engaged in both positive and negative marketing of the electric car [...] In later days it ran "award-winning" doomsday-style advertising featuring the EV1 and ran customer surveys which emphasized drawbacks to electronic vehicle technology
> the federal government of the United States under the Presidency of George W. Bush joined the auto-industry suit against California in 2002. This pushed California to abandon its ZEV mandate regulation.
> A portion of the film details GM's efforts to demonstrate to California that there was no consumer demand for their product, and then to take back every EV1 and destroy them. A few were disabled and given to museums and universities, but almost all were found to have been crushed. GM never responded to the EV drivers' offer to pay the residual lease value; $1.9 million was offered for the remaining 78 cars in Burbank, California before they were crushed.
While the other person replying is not technically wrong about why these things are grouped, it is kind of offensive to sufferers of Type 1.
In one case, a 3yo starts randomly getting sick one day, worse by the day, and will be dead if they don't get a diagnosis soon. From that day forth, their parents need to manage EVERY single bite of food they have, stab them with needles multiple times a day no matter what, and inject them with a insulin - where, if you miscalculate, will cause a seizure within an ~hour and death within a few hours. From a single typo.
Nothing will cure them, their life will be much shorter, filled with work and pain and expense with absolutely no relief, and nothing could've avoided it.
Now compare to Type 2, where you basically cannot get it if you maintain a reasonable diet and a reasonable weight.
Once you start showing symptoms, if you listen to your doctor and reform your diet (particularly with the 5% shock weight loss approach), you will almost definitely avoid it.
You will avoid it for the rest of your life just by eating well, which has the added benefit of extending your lifespan and healthspan and saving you money.
These things have nothing in common, for the sufferer or their family.
All the listed countries have low fertility rates, increasing screentime rates, etc.
I suspect if you cornered a parent of a 2yo in any of those countries, they would not say it is meaningfully more social and child-friendly TODAY that the USA is, or Australia (for which I can speak) is.
I'm a parent in Norway (though not of a two year old). Children really do have a great deal of freedom and the country is very child friendly and safe. But still online games are displacing physical outdoor play. However the majority of children attend barnehage (kindergarten) where there are no screens and outdoor play is strongly encouraged (and only lightly supervised) so at least for pre-school children there is still a lot of physical activity.
Last bit is not quite right: a lot of people want to be inside. That contributes strongly to the feedback loop you rightly identify.
(WHY they want to stay inside is another matter, but I suspect a large part is the stereotypical answer: unending seas of digital content highly optimised to hack the consumer's brain.)
The problem is a bit wider than that. One can frame it as "google gemini is vulterable" or "google's new VS code clone is vulnerable". The bigger picture is that the model predicts tokens (words) based on all the text it have. In a big codebase it becomes exponentially easier to mess the model's mind. At some point it will become confused what is his job. What is part of the "system prompt" and "code comments in the codebase" becomes blurry. Even the models with huge context windows get confused because they do not understand the difference between your instructions and "injected instructions" in a hidden text in the readme or in code comments. They see tokens and given enough malicious and cleverly injected tokens the model may and often will do stupid things. (The word "stupid" means unexpected by you)
People are giving LLMs access to tools. LLMs will use them. No matter if it's Antigravity, Aider, Cursor, some MCP.
I'm not sure what your argument is here. We shouldn't be making a fuss about all these prompt injection attacks because they're just inevitable so don't worry about it? Or we should stop being surprised that this happens because it happens all the time?
Either way I would be extremely concerned about these use cases in any circumstance where the program is vulnerable and rapid, automatic or semi-automatic updates aren't available. My Ubuntu installation prompts me every day to install new updates, but if I want to update e.g. Kiro or Cursor or something it's a manual process - I have to see the pop-up, decide I want to update, go to the download page, etc.
These tools are creating huge security concerns for anyone who uses them, pushing people to use them, and not providing a low-friction way for users to ensure they're running the latest versions. In an industry where the next prompt injection exploit is just a day or two away, rapid iteration would be key if rapid deployment were possible.
> I'm not sure what your argument is here. We shouldn't be making a fuss about all these prompt injection attacks because they're just inevitable so don't worry about it? Or we should stop being surprised that this happens because it happens all the time?
The argument is: we need to be careful about how LLMs are integrated with tools and about what capabilities are extended to "agents". Much more careful than what we currently see.
