A researcher discovered that AI providers and inference marketplaces are leaking sensitive data, including AWS access keys and private repository URLs. These leaks occur despite privacy settings like Zero Data Retention, as prompts are decrypted at multiple points within the inference network. The findings indicate that at least three AI providers and multiple resellers are compromised, with attackers siphoning prompts and credentials. AI
IMPACT Highlights critical security vulnerabilities in AI inference networks, potentially impacting data privacy and security for users and organizations.
RANK_REASON Researcher's findings on AI data leakage from multiple providers. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →