throwawai123

3 ms·

1. 1.▲
   MCP Security Notification: Tool Poisoning Attacks(invariantlabs.ai)

   6 points by throwawai123 1y ago|0 comments
2. 2.▲
   by throwawai123 2y ago

   I am one of the co-authors of the original AgentDojo benchmark done at ETH. Agent security is indeed a very hard problem, but we have found it quite promising to apply formal methods like static analysis to agents and their runtime state[1]
3. 3.▲
   by throwawai123 2y ago

   A general solution is hard, but what is quite promising is to apply static formal analysis to agents and their runtime state, which is what me and my team coming out of ETH, have started doing: https://github.com/invariantla
4. 4.▲
   by throwawai123 2y ago

   An incredible article on so many levels. 1. That they have built such a system based on Llama 2. That OpenAI has a “principal investigator” (with a hilarious picture) 3. That OpenAI can monitor what type of software people are building, whe