If you enjoy this newsletter, please become our paid subscriber to help this keep going.
Highlights
Rethink IAM as AI agents.Amazon: AI-generated alert.OWASP AI testing guide.US Retail Data Breach.Anthropic’s AI weird experiment.
Special!
Deep Dive
1. Rethink IAM as AI agents Venture Beat
80% breach: Stolen credentials
Difficult to manage AI agents.
Now use Bluetooth Low Energy
Cisco’s Duo, Microsoft Entra ID, Ping Identity.
Embrace “Never trust, always verify”.
2. Amazon: AI-generated Alert Amazon
For rind doorbell division.
Alert unsual activities around homes.
25/6: Beta version for premium users.
Plans to push deeper into AI.
Concern about flaws with the technologies.
3. OWASP AI Testing guide OWASP
Comprehensive methodologies, best practices.
Principles and objectives to test AI.
Threat modelling for AI systems..
Application, model, infrastructure, data.
Open source project.
4. US Retail Data Breach HowToGeek
2.2 millions of records,
Sensitive personal, finance and health info.
Current and former employees.
Global Food Retail Company.
First accesss: November 6, 2024.
5. Anthropic’s AI Weird Experiment. TechCrunch
I vending machine went haywire.
Hallucinated, lied, & contacted security.
Obsessed over selling tungsten cubes.
Long-running instance caused odd behavior.
AI middle-managers still seem possible.
Our upcoming book
Notice: Y2 GenAI Safety and Security is on GumRoad and Amazon with paperback.
Share this post