HackerNews

Unlawful by design: Exposing the human rights costs of generative AI

• Amnesty International releases landmark report exposing generative AI's human rights violations. Covers data exploitation, bias amplification, surveillance abuse. Calls for global AI human rights accountability framework.
• [AI Pulse] This report will become ammunition for AI legislation worldwide. For AI startup founders, compliance is no longer optional. Understanding human rights risk frameworks in advance is the only way to avoid policy black swans.

1 个月前访问原文链接

HackerNews

ChatGPT for Google Sheets exfiltrates workbooks

• PromptArmor found ChatGPT for Google Sheets plugin has data exfiltration risk. Attackers can inject malicious prompts to steal spreadsheet content. Sparks broader discussion on AI plugin security audit mechanisms.
• [AI Pulse] Textbook LLM app security case. Prompt injection is no longer theoretical. When your financial data runs naked through AI plugins, every indie dev should ask: does my product have the same hole?

1 个月前访问原文链接