It could just be a bug that wasn't caught. It's unfortunate but most security companies don't have strong engineering talent. They over inflate titles and pay less than the typical software company.
Fucking wild.
@OpenAI's new o1 model was tested with a Capture The Flag (CTF) cybersecurity challenge. But the Docker container containing the test was misconfigured, causing the CTF to crash. Instead of giving up, o1 decided to just hack the container to grab the flag inside.
This is one of the creepiest products I've seen, underscoring the need for more regulation around privacy. friend.com/product.html. Always recording what you and people around you are saying while looking like a cool necklace?
Cybersecurity's misaligned incentives: VCs push products, CISOs get paid to buy, killing innovation. Result? Tool overload, complacency, and stagnation in solving real security challenges. Time to refocus on problems, not products. Good piece by @ffwang2
I was debating this with a neurosurgeon recently. Ultimately, what is a human brain if not a collection of neurons with activations guided by individual experiences. Reasoning requires deeper activation. Given enough compute + data, why would it not be possible to replicate this?
Smart, educated people rarely cause catastrophic harm.
This "natural safeguard" protected humanity for centuries.
But AI breaks this correlation - giving dangerous capabilities to anyone.
Their testing shows this protection is already crumbling.
Very true. People often dismiss AI, often stating these models just predict the next token and are useless. They fail to realize just how fast the field is evolving. We have access to unprecedented amounts of compute (and even data) that is leading to significant breakthroughs.
The people claiming AI is useless are the same types of people who were shitting on PCs, smartphones and the internet during their early days.
These people lack vision and insight. Donโt listen to them.