Oktane Preview: building frameworks to secure our Agentic AI future
Like it or not, Agentic AI and protocols like MCP and A2A are getting pushed as the glue to take business process automation to the next level. Giving agents the power and access they need to accomplish these lofty goals is going to be challenging, from a security perspective.
How do put AI agents in the position to perform broad tasks autonomously without granting them all the privileges? How do we avoid making AI agents a gold mine for attackers - the first place they stop once they hack into our companies? These are some examples of the questions Okta aims to answer at this year’s Oktane event, and we aim to kick off the conversations a little early - with this interview!
Segment Resources:
Reports of indirect prompt injection issues have been around for a while. Of particular note was Michael Bargury's Living off Microsoft Copilot presentation from Black Hat USA 2024. Simply sending an email to a Copilot user could make bad stuff happen.
Now, at Black Hat 2025, we've got more: the ability to plunder any data resource connected to ChatGPT (they call these integrations "Connectors") from Tamir Ishay Sharbat at Zenity Labs. The research is titled AgentFlayer: ChatGPT Connectors 0click Attack.
Looks like Google Jules is also vulnerable to what the Embrace the Red blog is calling invisible prompts. Sourcegraph's Amp Code is also vulnerable to the same attack, which encodes instructions to make them invisible.
What's really going to ruffle feathers is the fact that all these companies know this stuff is possible, but don't seem to be able to figure out how to prevent it. Ideally, we'd want to be able to distinguish between intended instruction and instructions injected via attachments or some other means outside of the prompt box. I guess that's easier said than done?
Finally, in the enterprise security news,
All that and more, on this episode of Enterprise Security Weekly.
Show Notes: https://securityweekly.com/esw-421
Vulnerability management is broken. Organizations basically use math to turn a crappy list into a slightly less crappy list, and the hardest part of the job as a CIO is deciding what NOT to fix. There has to be a better way, and there is...
Segment Resources:
This segment is sponsored by Horizon3.ai. Visit https://securityweekly.com/horizon3 to learn more about them!
Andy Ellis visited every booth at Black Hat. Every. Single. One. He wrote up what he learned and we discuss his findings!
https://www.duha.co/state-of-security-vendors-blackhat-2025/
Finally, in the enterprise security news,
All that and more, on this episode of Enterprise Security Weekly.
Show Notes: https://securityweekly.com/esw-420
We're coming live from hacker summer camp 2025, so it seemed appropriate to share what we've seen and heard so far at this year's event. Adrian's on vacation, so this episode is featuring Jackie McGuire and Ayman Elsawah!
Then, in the enterprise security news,
All that and more, on this episode of Enterprise Security Weekly.
Show Notes: https://securityweekly.com/esw-419
This week, we’ve had to make some last minute adjustments, so we’re going to do the news first, split into two segments.
This week, we’re discussing:
Guillaume shares his experiences building security from scratch at Canadian FinTech, Finaptic. Imagine the situation: you're CISO, and literally NOTHING is in place yet. No policies, no controls, no GRC processes. Where do you start? What do you do first? Are there things you can get away with that would be impossible in older, well-established financial firms?
Show Notes: https://securityweekly.com/esw-418