Fun stuff from my team mates Rene Rehme @renereh1, Nina Piontek and @kantorkel:
-
Fun stuff from my team mates Rene Rehme @renereh1, Nina Piontek and @kantorkel:
"Via a large enterprise's AI assistant, we obtained access to several million Entra identities and all chat logs including attachments — no prompt injection or model tricks required.”
We don't need to hack your AI Agent to hack your AI Agent - SRLabs Research
We strolled through an enterprise AI assistant's backend, helped ourselves to full application takeover and access to every chat log, and had a Microsoft Entra ID dump for dessert — no prompt injection, no model tricks, no AI expertise required.
SRLabs (srlabs.de)
Security Research Labs (@srlabs@infosec.exchange)
We don't need to hack your AI Agent to hack your AI Agent …and we don't need an AI agent for that either :) Via a large enterprise's AI assistant, we obtained access to several million Entra identities and all chat logs including attachments — no prompt injection or model tricks required. For all we know, the poor agent was not at fault and may not have even been able to witness what was happening. https://srlabs.de/blog/hacking-ai-agent #AI #AIhacking #VulnerabilityDisclosure #ResponsibleDisclosure
Infosec Exchange (infosec.exchange)
-
Fun stuff from my team mates Rene Rehme @renereh1, Nina Piontek and @kantorkel:
"Via a large enterprise's AI assistant, we obtained access to several million Entra identities and all chat logs including attachments — no prompt injection or model tricks required.”
We don't need to hack your AI Agent to hack your AI Agent - SRLabs Research
We strolled through an enterprise AI assistant's backend, helped ourselves to full application takeover and access to every chat log, and had a Microsoft Entra ID dump for dessert — no prompt injection, no model tricks, no AI expertise required.
SRLabs (srlabs.de)
Security Research Labs (@srlabs@infosec.exchange)
We don't need to hack your AI Agent to hack your AI Agent …and we don't need an AI agent for that either :) Via a large enterprise's AI assistant, we obtained access to several million Entra identities and all chat logs including attachments — no prompt injection or model tricks required. For all we know, the poor agent was not at fault and may not have even been able to witness what was happening. https://srlabs.de/blog/hacking-ai-agent #AI #AIhacking #VulnerabilityDisclosure #ResponsibleDisclosure
Infosec Exchange (infosec.exchange)
@linuzifer @renereh1 @kantorkel django debug mode in owasp top 10 when?
-
R relay@relay.infosec.exchange shared this topic
