Permit AI Access Control - Fine-Grained Permissions for AI-Powered Applications
by•
Permit.io AI Access Control brings fine-grained authorization (FGA) to AI workflows, ensuring AI models interact safely with sensitive data, external APIs, and users - all without developers having to build this from scratch
Replies
Andsend
Way to go Gabriel L. Manor
Permit.io
Thanks for the kind words,@aage_reerslev1 !
Amazing concept - definitely a time saver!
Permit.io
@skail looking forward to hear your experience with trying it 😉
Really impressed with Permit's AI Access Control integration. From the Langchain integration docs, it is straightforward to implement fine-grained permissions in AI applications without the security headaches. The JWT validation tools are particularly useful for authenticating requests to AI services.
Curious - are there plans to expand the retrievers to support more vector database integrations beyond FAISS? Or can it be combined with other vector store like BM25?
Permit.io
Thanks for the kind words, @taofiq_aiyelabegan !
To your question: YES AND YES! Permit's LangChain retrievers built in a way that made it pluggable to every vector store in LangChain. Looking forward to see you in the platform, and hear your detailed review on it!
DiffSense
Any false negatives with a system like this, and if so how are you mitigating it to avoid disruption of service / user experience? I.e: When im working with OpenAI. Sometimes it kicks me out because I thinks im doing something im not suppose to. But all I do is work on code that is security related etc.
Permit.io
Hey, @sentry_co ,
We tried to build our integrations in a way that puts the most structured policy checks in the unstructured world of LLM..
Actually what we (and trending frameworks like MCP and PydanticAI) is trying to do, is avoid the exact same mistakes that you describe here
Ovren
Congrats on the launch of Permit AI Access Control! It's great to see a solution that addresses the critical need for secure and compliant AI workflows. This will definitely help developers ensure their models interact safely with sensitive data and external tools!
Permit.io
Thanks, @mikita_aliaksandrovich!
Looking forward to see you in the platform!
That’s awesome team Permit!!🥳🔥👾
Congratulations to @or_weis and his team. 🎉
Congrats! This is certainly a need as we go even further into AI/ML use cases
Awesome news. I have been following Permit since August 2023 and its cool to see how they are modernizing user permissions with AI