Discussion – has anyone build a firewall for AI models yet?
2•Akhiljp•1h ago
Trying to figure out if there are already companies that have build firewall like products for AI models. Assuming everyone will now start hosting open source models to control their destiny, I wonder if this is a large market for startups to go after.
Comments
KynaraAI•1h ago
I am not a firewall or security expert but would you have organizations protect their sensitive data from AI's reach thereby limiting AI's ability ? Wouldn't the real ROI of AI will depend on the sensitive data ?
I guess dynamic firewall's with runtime permissions are needed protect the data and at the same time let AI do it's thing.
Akhiljp•59m ago
I beg to differ - manipulating a model to extract data is worse that protecting sensitive data. I am curious as to why this is not a big space yet?
verdverm•57m ago
We already have many components and solutions for "firewalls", like everything we've done with microservice isolation. If you mean "sandbox" more than "firewall", there are 100s since Claude Code came out
Akhiljp•50m ago
So my first thesis that I want to prove is - are all enterprises going to start self hosting open source models ? If yes, then one will need to deploy a solution around the models to act as a firewall. A firewall fine tuned for the context that’s coming in and coming out of a model. Way different from how microservices work today.
verdverm•3m ago
1. No, most enterprises are not self hosting at this point, hardware is too expensive. Eg. we are buying Fireworks.ai
2. "firewall" is the wrong term, you are looking for "guardrails" if you are talking about prompt protections.
pagutierrezn•22m ago
When you say firewalls for AI models, you mean firewalls for agents, right? Otherwise, I don't see the need to firewall a model.
Akhiljp•15m ago
I meant for a model. Like agents could be many within an organization, instead of patching up each agent with a security solution, wouldn’t it be easier to build one that wraps around the model ?
KynaraAI•1h ago
I guess dynamic firewall's with runtime permissions are needed protect the data and at the same time let AI do it's thing.
Akhiljp•59m ago
verdverm•57m ago
Akhiljp•50m ago
verdverm•3m ago
2. "firewall" is the wrong term, you are looking for "guardrails" if you are talking about prompt protections.