neapolisbeach

Joined 8 years ago

Karma

425

Submissions

39

Member since

Nov 26

View on Hacker News

Recent submissions

Comment · 9 months ago

https://hmcguinn.com/

Natural emergent misalignment from reward hacking in production rl [pdf] assets.anthropic.com

3 neapolisbeach 11 months ago 0 comments

Whisper leak: a side-channel attack on large language models arxiv.org

3 neapolisbeach 11 months ago 0 comments

Google Private AI Compute [pdf] services.google.com

1 neapolisbeach 11 months ago 0 comments

Comment · 1 year ago

Nsjail, firecracker, gVisor, or v8 isolates are all good options with different tradeoffs

Comment · 1 year ago

This comment reads as 100% LLM generated

Boeing whistleblower found dead in US bbc.com

1,335 neapolisbeach 2 years ago 556 comments

Finding a command injection vulnerability in my smart thermostat (CVE-2023-4212) hmcguinn.com

2 neapolisbeach 3 years ago 0 comments

Vm2 discontinued due to unfixable security issues github.com

2 neapolisbeach 3 years ago 0 comments

Comment · 3 years ago

If you're interested in time-lock puzzles, drand [1] achieves this but through randomness instead of computational effort.

[1] https://drand.love/

Comment · 3 years ago

Looks like a 404, would love to see the gist though!

Comment · 3 years ago

Comment · 4 years ago

Port knocking maybe?

https://en.m.wikipedia.org/wiki/Port_knocking

Comment · 4 years ago

I built a similar scheme for a class project, but relying on verifiable delay functions [1] to provide the time lock.

Might be something you’re interested in.

[1] https://eprint.iacr.org/2018/601.pdf