If you enjoy our newsletter, please consider to be a paid subscriber to help us keep more news and updates coming out.
Highlights
OpenAI confirms Breach.US Banks data theft.Google new AI coding tool get hacked.Petri: Open-source auditing tool.Google Nano Banana Pro Fake IDs.
Deep Dive
OpenAI confirms Breach Decrypt
A breach at MixPanel on Nov 8.
Attackers gained access to systems.
Exported dataset of customers and analytics.
Exposed names, email addresses, and browser locations.
Could be used for phishing attempts.
US Banks access data theft TechCrunch
SitusAMC confirms a breach on Nov 12.
Hackers stole “accounting records and legal agreements”.
SitusAMC processes billions of documents related to loans annually.
Scope and nature of attacks remain under investigation.
Notifications sent to JPMorgen Chase, Citigroup and Morgen Stanley.
Google new AI coding tool get hacked Forbes
Flaw found within 24 hours.
Hackers can install malware.
Creates persistent system backdoor.
Affects both Windows and Mac.
No patch currently available.
Reinstalling does not fix it.
Petri: Open-source auditing tool Anthropic
Accelerates AI safety research.
Automates model behavior testing.
Agents simulate user interactions.
Judges score transcript safety.
Scales beyond manual auditing.
Tested fourteen frontier models.
Sonnet 4.5 performed best.
Analyzes deception and power-seeking.
Code available on GitHub.
Google Nano Banana Pro Fake IDs OECD
Generate highly realistic fake Identity documents.
Can bypass legacy verification systems.
Concerns about ID theft and fraud.
Happens to various types of documents.
Indian Aadhaar & PAN cards, US licenses, passports.
![[Available] Book Report Q3, 2025](https://substackcdn.com/image/fetch/$s_!HI5v!,w_140,h_140,c_fill,f_webp,q_auto:good,fl_progressive:steep,g_auto/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F62b0655a-9a73-4382-8201-d9007269e7ad_900x900.jpeg)
