AI giants' shared encryption keys exposed user data, researchers find

Researchers discovered that major AI companies, including OpenAI, Google, Meta, and Microsoft, used identical encryption keys for all users. This practice allowed them to access the internal "thoughts" of AI models. The study, involving institutions like Stanford University and UC Berkeley, found hundreds of active passwords and API keys within public logs, exposing sensitive data. AI

IMPACT Exposes critical security flaws in how leading AI models handle user data, potentially impacting trust and requiring immediate industry-wide security updates.

RANK_REASON Research paper detailing a security vulnerability in AI model encryption. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — mastodon.social →

• Carnegie Mellon University
• Claude 3
• GPT-4
• Llama 2
• Mastodon
• Meta
• Microsoft
• OpenAI
• Stanford University
• University of California, Berkeley
• University of Washington

• safety
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →