Can anyone recommend application security best practices for generative AI tools?


2.8k views4 Comments

Chief Information Security Officer in Software, 11 - 50 employees
I would say quite the same as a standard application with an additional focus on data exposure. My main focus when my company started using Generative AI was focus on data ingested by the AI (we are an EU based company), and also the security of the Generative AI provider.
2 1 Reply
Information and Security Office & Enterprise Data Governance/AI in Finance (non-banking), 1,001 - 5,000 employees

I agree with  that the rules of engagement do not change with AI in the picture; you still have to follow the same rules of DevSecOps and have the same level of due diligence to ensure your code is designed to provide that is designed to do, including performing needed Threat Modeling, Design Reviews, keeping in mind transparency and accountability with Privacy by design and Security by Design.

1
VP Information Security Assurance, 10,001+ employees
There are the following dimensions in my mind
1) controls like
     a) Input validation (while maintaining the spirit of natural language). you dont want your LLMs to crash / or elevate privilege
    b) ensure relevant privacy conditions are built in specially when the model attempts to store the questions as an input for its future learning. So while the user would appreciate the result LLMs will thru s/he may need his own data anonymized in results
   c) boundary conditions such that queries or its results dont overwhelm the environment and make service unavailable

2) "intelligence in response" such that the LLMs are not fooled in providing responses that might be counter protective.for example. "how to hack LLMs" may get no result but question like " is there current weakness that the LLM is slf healing" might, giving important reckon

3) the LLMs itself, like immutable logs, may need to be protected from tampering 
4) protect the knowledge set from where it currently responds based on incremental context learning. so that it doesn't get poisoned 

OWASP has a reference for LLMS OWASP Top 10 for Large Language Model Applications | OWASP Foundation, check it out pl

Keen to learn your perspectives when possible pl
1
CTO in Consumer Goods, 11 - 50 employees
Human in the loop. Exercise a lot of caution around agent applications that have integrations beyond information retrieval. eg database updates, scoring, api updates, etc.
2

Content you might like

Data breaches due to remote work27%

Ransomware attacks43%

Lack of a corporate security plan13%

Missing security patches8%

Failure to inform employees of threats3%

Other (please specify)3%


349 PARTICIPANTS

846 views1 Comment

Yes, most security leaders.37%

Yes, some security leaders.53%

No7%

Not sure2%


355 PARTICIPANTS

987 views1 Comment

CTO in Software, 201 - 500 employees
Without a doubt - Technical Debt! It's a ball and chain that creates an ever increasing drag on any organization, stifles innovation, and prevents transformation.
Read More Comments
46.5k views133 Upvotes324 Comments