F5 Labs Threat Research team analyzed Morality Dilemma jailbreak, where attackers feed corrupted ethical rules to AI models to bypass guardrails.
Public source
Publisher name
Public post
Could your AI model’s strength be its biggest vulnerability? 🤖🛡️ The F5 Labs Threat Research team takes a closer look at Morality Dilemma jailbreak; attackers are succ…
Company
F5
We deliver and secure every app.
- Industry
- IT Services and IT Consulting
- Location
- Seattle, US
- Company size
- 5,001–10,000 employees
About F5
F5, Inc. (NASDAQ: FFIV) is the global leader in delivering and securing every app and API. As hybrid multicloud increases complexity, AI inference changes application architectures and traffic patterns, and the threat landscape expands, F5 converges application delivery and security through the F5 Application Delivery and Security Platform (ADSP) to keep applications and APIs fast, available, and secure wherever they run, on premises, in the cloud, at the edge, and across hybrid, multicloud environments. For more than 30 years, F5 has kept mission-critical applications running for the world's largest organizations. Today, F5 extends that leadership for the AI era, partnering with enterprises to deliver the performance, security, and reliability modern business requires. Together, we help each other thrive and build a better digital world.
See moreLatest activity
Latest activity from F5
123 signals
Presence & Recognition
F5 is attending the AI Infra Summit in Santa Clara Convention Center next week to discuss intelligent routing and runtime protection at the cluster border.
Products & Services
F5 announced that the next evolution in enterprise AI security is model routing, featuring specialized models, policy-based routing, and local endpoints.
Presence & Recognition
F5 Chief Marketing Officer John Maddison sat down with SiliconANGLE & theCUBE at CrowdStrike Fal.Con last week to discuss AI-powered security and zero-day attacks.
Discover more
Similar signals
Similar public activity from other companies.
Research & Knowledge
Adapture
Adapture published a blog exploring how F5 AI Guardrails and F5 AI Red Team capabilities help organizations strengthen AI security, improve governance, and build confidence in AI deployments.
Research & Knowledge
BlueArmor, LLC
BlueArmor, LLC published a new analysis of AI-generated apps testing 434 exploitable security flaws across the tested applications.
Research & Knowledge
Zero Trust Cyber & AI Advisors
Zero Trust Cyber & AI Advisors reports that Booz Allen tested 18 advanced U.S. and Chinese AI models and found that a leading frontier model autonomously executed an end-to-end network intrusion without human guidance.
Research & Knowledge
Radware
Radware published research showing how prompt injection can turn AI into a phisher.
Research & Knowledge
Grip Security