GenAI Safety & Security | June 30 - July 6, 2025
If you enjoy this newsletter, please become our paid subscriber to help this keep going.
$180M stolen from a bank.Quantas: 6 millions profile breach.Luis Vuitton customer breach.US to develop AI security playbook.Red teaming for AI safety.
Special!
Exploit a vulnerabilities in the payment network.
Drained reserve accounts.
Moved the fund through Bitcoin and USDT.
Crypto platform freezed suspicious transaction.
Returned some block funds.
Third party customer service platform.
30/6: Detect unusual activities.
Data: names, email, address, phones, BOD.
Also include Frequent Flyer Numbers.
Still under investigation in Australia.
June: System breach .
Include Contact information.
Did not include financial information.
May 2025: Investigation for 2024 breach.
Include Christian Dior Couture and Tiffany.
Called: "Advanced AI Security Readiness Act,"
To address vulnerablities, threat detection.
Cyber and physical strategies and plans for AIs.
Highlighting US-China AI competition.
Cited DeepSeek and how GPUs were smuggled.
Attempt to break AI system to reveal weakness.
Red Team simulates an adversarial effort.
How bad this AI go wrong or made to do something bad.
Then systematically test those scenarios.
Examine a whole pipeline (data, infrastructure UI, etc)
Thanks for reading Secure GenAI ! This post is public so feel free to share it.
Notice: Y2 GenAI Safety and Security is on GumRoad and Amazon with paperback.
