

This is just a standard prompt hack. This will always exist with llms. They don’t have any real understanding of language so safety protocols can’t actually ban topics, only sets of words and phrases.
There was an extensive set of prompts working toward elder abuse before the result in question.
My guess is that the redditor who discovered it disguised it to look like homework and reproduced the hack, and added the “brother” to create more authentic rage bait.
It’s a combination of issues. First is compatibility issues. Like logging in on mobile web or app with a passkey doesn’t reliably work for me. It might have been due to the password manager, but for some things the option wasn’t even there afaict. If I’m going to really switch to passkeys, I want it to work more reliably.
The second is usability. Passwords in a password manager are a 2 click entry on the username or password form field. Password managers have streamlined this system over the past decade.
Passkeys, ironically, required more steps when pulling from the password manager, including required clicks in less convenient places. I hope these types of issues get ironed out eventually.
Someone probably knows the answer. Ask around. This should include either customers or customer advocates. If nobody knows the answer, then do the simplest thing that accomplishes what you need in order to proceed. Sometimes that means doing nothing. If there are multiple ways to accomplish what you need, do the one that leaves you in a more flexible state for future changes. You can bring up your choices or decisions to team members if you need, possibly during a standup or just ad-hoc.
If you aren’t empowered to take one of those steps, then you are in a dysfunctional environment, in that case, collect your salary and keep your head down, and if you are so inclined, try to find a new company or team to join.
Yeah for sure. And regardless, I know some great security engineers who are ADHD.