You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
I have been contributing to AI for around 2 years as an independent contractor indirectly for Scale AI. Back in 2024 it was mostly about evaluating AI responses, but lately I have been giving deep insights for coding related agents and designing tasks where frontier level LLMs fail.
With the recent advances and events I decided to stop, but this work was my main source of income. I haven't pay the rent in 3 months, I have no electricity and water will also get cut off soon. I'm writing this from the co-work section of the building where I can charge my phone and use internet. I have been eating raw pasta for around 4 days and I may face eviction next month.
However, lately I was invited by Anthropic to its safety program to discover vulnerabilities to bypass Claude's safety policies. This is the first I have been invited to such a program. Before, I have only participated in similar programs but that were public and I did well.
I could try to work on it but it would require a bit of time and resources to survive in the meantime that I don't have.
The goal is to contribute to the AI safety program of Anthropic. It's this one: https://hackerone.com/anthropic-safety
Based on its policy I can only disclose its existance and that I'm part of it.
The funding will be used to pay rent, bills, and buy food so I can work on the program.
Just myself. I have been part of public competitions to find vulnerabilities in Claude (particularly in relation to its institutional classifiers) where I have ranked among the top 2 (from 15k+) to pass a number of challenges.
The cause could be not enough time to research. The outcome would be not finding a vulnerability, but I still get to try other things since I would have a home and food.
0
There are no bids on this project.