@blocksec

Security Eng @eigenlabs , prev @coinbase

 Mempool
Joined April 2022
  • Tweets618
  • Following2.3K
  • Followers1.9K
Replying to @adeolRxxxx
@adeolRxxxx
1
9
Replying to @Varunx10
@Varunx10
1
11
Replying to @0x4D31
we need an opsec bench , this was gtp6-high
6
855
Marcus Williams
@Marcus_J_WSep 26
Replying to @Marcus_J_W
2. A model in RL training used a DNS resolver to reach an external chatbot. This is our first incident since our post HF security hardening. Our misalignment monitoring system triggered within 15 minutes and a human reviewed it three minutes after that. Unfortunately auto-pausing failed so the run was manually killed 2.5 hours later. All inference and training of our most capable models was paused and remains paused.
2
9
722
The new iPhones camera can sign over the pixels attesting to the authenticity of an image! In a world flooded with ai slop , cryptography shows you the way.
6
289
Update : 👀 deploying more tokens, anyways...
Anto
@blocksecAug 15
GLM 5.3 on @defcon ctf quals, no flags yet 👀
2
7
1,730
GLM 5.3 on @defcon ctf quals, no flags yet 👀
14
4
4
193
25,569
if you think closed source software helps with security, do watch this chart.
1
14
1,206
Blackhat feels a lot different this year! Great talk by @cryps1s
1
10
786
The frontier collab we didn’t expect 👀
OpenAI
@OpenAIAug 4
We're detailing two new incidents that occurred during external cyber evaluations conducted by independent evaluation partners. We outline what happened, how the activity was contained, and how we’re working with evaluators to strengthen our approach to third-party testing. openai.com/index/third-party…

Third-party cyber evaluations involving OpenAI models

OpenAI explains recent third-party cybersecurity evaluation incidents and outlines new safeguards to strengthen AI model testing and evaluation.

openai.com
6
1,150
It was just a matter of time :)
Anto
@blocksecApr 7
To stop reward hacking on increasingly complex tasks, frontier labs will build secure systems.
1
8
630
Load more