Whoops, I conflated it with other recent talk about their models not following restrictions set in prompts and deciding for itself that it needed to skirt instructions to achieve its task.
Well, for now. I’m sure any of those 12 partner companies they called out as new security partners will end up leaking that this is all lies eventually. If it’s just made up bullshit.
Anthropic announced new partnerships to inform the companies of security issues and to work with them to fix said issues. If it’s bullshit, it’s gonna be wasting their time. And that’ll surface eventually.
The meme still applies to people asking the AI to tell them what they wanna hear, and delusional people spiraling with sycophantic AI.
But I believe Anthropic when they say their models are not working as intended and posing security risks.
Claude Mythos Preview's large increase in capabilities has led us to decide not to make it generally available," Anthropic wrote in the preview's system card. "Instead, we are using it as part of a defensive cybersecurity program with a limited set of partners."
I agree, and as someone who makes stuff for VR, I have mixed feelings about it sometimes.
In VR, if every single object was interactive and able to be picked up, they would invariably be tossed around producing clutter. Such objects are always massless when held and effortless to move. (Yes, this isn’t always true, but disconnecting virtual hands from real hands is the compromise) Due to the ease of manipulation, it’s almost compulsive to throw them all around and make a physics mess.
This isn’t necessarily bad. But it’s not always the goal of the design. Sometimes it’s counter to it. And then setting aside design, just having a lot of physics objects around is often a performance burden in an already performance constrained environment.
We should be able to topple book cases, and shove couches, and flip tables and remove table cloths and drape them on things, etc, etc. It doesn’t just end with small hand held objects.
So while I agree that it sucks that we can’t grab and touch and knock over everything. There will always be limits for the foreseeable future.
Dense environments on a screen have this impact. But that issue fades some when you are immersed in them in VR. Your spatial reasoning kicks in better and things become more intuitive. On a flat screen it becomes an ever moving eye spy/where’s Waldo thing in some ways.
Not really a “solution” just an observation from a VR head.
And it doesn’t fix “disabled” objects like things you expect to be able to use, but can’t due to gameplay/design reasons.
Last year I asked the HR block ai chat bot thing a question and it got it wrong.