Suggestions
Mark Pundsack
CEO @ Styra, Ex-Replicated, Ex-GitLab, Ex-Heroku. Advisor and angel investor to DevOps startups.
Mark Pundsack is the Chief Executive Officer (CEO) of Styra, a company recognized for its work on Open Policy Agent (OPA) and focused on reinventing authorization for cloud-native applications. He was appointed to this role in April 2023, succeeding in leading the company's product strategy and development efforts.245
Background and Experience
Professional Journey:
Before joining Styra, Pundsack had a significant impact at GitLab, where he contributed to the company's growth during its early years. His expertise lies in product strategy and DevOps, which he now leverages to advance Styra's mission of providing unified solutions for authorization challenges across various cloud environments.34
Leadership Role:
As CEO, Pundsack is tasked with steering Styra's initiatives to enhance policy as code and improve enterprise-level authorization solutions. His leadership is expected to drive innovation and address the complexities associated with modern cloud architectures.125
Vision for Styra
Pundsack's vision focuses on addressing the "entitlement explosion" that organizations face as they increasingly adopt cloud-native technologies. By implementing robust authorization frameworks, he aims to ensure that enterprises can manage access control effectively while maintaining security and compliance.13
In summary, Mark Pundsack's extensive experience in product strategy and DevOps positions him as a pivotal figure in shaping the future of authorization solutions at Styra.
Highlights
Enterprise AI adopters, did you miss the "AI Risk Management Framework: Generative AI Profile" just released by the US National Institute of Standards and Technology (NIST)?
As AI becomes more complex and less controllable, the right access controls become all the more important to mitigate risks such as backdoors, leakage, data poisoning, and unauthorized use. Because emergent AI behavior is impossible to fully control, safeguarding both training and production data are key to an organization's information security.
"Information security for GAI models and systems also includes security, confidentiality, and integrity of the GAI training data, code, and model weights. [A novel] cybersecurity risk to GAI is data poisoning, in which an adversary compromises a training dataset used by a model to manipulate its operation."
The NIST document also issues a broad call-to-action to protect privacy in training and production data.
"Establish [...] access controls for training sets or production data containing biometric, confidential, copyrighted, licensed, patented, personal, proprietary, sensitive, or trade-marked information, in accordance with privacy and data governance policies and data privacy laws."
While the call for access control is straightforward, its real-world implementation is anything but. How does your organization implement access control policies to safeguard your AI operations without slowing time-to-market?
#ai #generativeAI #NIST #accesscontrol
Recently, the OWASP Top 10 came out for LLM (#GenAI) applications. It’s interesting to see 4 of the top 10 are related to access control.
What’s most important for your organization?
