Comet AI browser can get prompt injected from any site, drain your bank account

Must we learn the same lessons over and over again? Why? Is our industry particularly stupid? Or just lazy?

Rather: it's perpetually in a rush for business reasons, and concerned with convenience. Security generally impedes both.

The winner (financially, and DAU-wise) is not going to be the one that moves slowly because they are building a secure product. That is, you only need security when you are big enough to either have Big Business customers or big enough to be the target of lawsuits.

Financially motivated to not prioritize security.

It's hard to sell what your product specifically can't do, while your competitors are spending their time building out what they can do. Beloved products can make a whole lot of serious mistakes before the public will actually turn on them.

LLMs can't learn lessons, you see, short context window.

"Our bridges don't collapse" is a selling point for an engineering firm, on something that their products don't do.

We need to stop calling ourselves engineers when we act like garage tinkerers.

Or, we need to actually regulate software that can have devastating failure modes such as "emptying your bank account" so that companies selling software to the public (directly or indirectly) cannot externalize the costs of their software architecture decisions.

Simply prohibiting disclaimer of liability in commercial software licenses might be enough.

Information security is, fundamentally, a misalignment of expected capabilities with new technologies.

There is literally no way a new technology can be "secure" until it has existed in the public zeitgeist for long enough that the general public has an intuitive feel for its capabilities and limitations.

Yes, when you release a new product, you can ensure that its functionality aligns with expectations from other products in the industry, or analogous products that people are already using. You can make design choices where a user has to slowly expose themselves to more functionality as they understand the technology deeper, but each step of the way is going to expose them to additional threats that they might not fully understand.

Security is that journey. You can just release a product using a brand new technology that's "secure" right out of the gate.

+1

And if you tried it wouldn’t be usable, and you’d probably get the threat model wrong anyway.

Call yourself whatever you choose, but the garage tinkerers will always move faster and discover new markets before the Very Serious Engineers have completed the third review of the comprehensive threat model with all stakeholders.

Yes, they will move fast and they will brake things, and some of those breakages will have catastrophic consequences, and then they can go "whoopsy daisy", face no consequences, and try the same thing again. Very normal, extremely sane way to structure society

Nobody cares about bridges collapsing if you built the first bridges and none have collapsed yet from the couple first folks trying them out, though.

It's only when someone tries to drive their loaded ox-driven cart through for the first time that you might find out what the max load of your bridge is.

its a steady stream of naive start ups run by people who think starting a company is something you do at the beginning of your career with no experience vs the end of your career with decades of experience.

I'm sorry but that's a pathetic excuse for what's going on here. These aren't some unpredictable novel threats that nobody could've reasonably seen coming.

Everyone who has their head screwed on right could tell you that this is an awful idea, for precisely these reasons, and we've known it for years. Maybe not their users if they haven't been exposed to LLMs to that degree, but certainly anyone who worked on this product should've known better, and if they didn't, then my opinion of this entire industry just fell through the floor.

This is tantamount to using SQL escaping instead of prepared statements in 2025. Except there's no equivalent to prepared statements in LLMs, so we know that mixing sensitive data with untrusted data shouldn't be done until we have the technical means to do it safely.

Doing it anyway when we've known about these risks for years is just negligence, and trying to use it as an excuse in 2025 points at total incompetence and indifference towards user safety.

The only reason this works out the way it does is because certain governments have been corrupted by business interests to the point that businesses don't have to face any accountability for the harm that they cause.

If companies were fined serious amounts of money and the people responsible went to prison if they committed gross negligence and harmed millions of people, the attitude would quickly change. But as things stand, the system optimizes for carelessness, indifference towards harm, and sociopathy.