AI Risks September 20, 2025 · The Register

OpenAI plugs ShadowLeak bug in ChatGPT that let miscreants raid inboxes - The Register

"ShadowLeak weaponizes the very capabilities that make AI assistants useful: email access, tool use, and autonomous web calls," Radware ...

Read the original at The Register ↗

Topics

Shadow AI

Mentions

OpenAIChatGPT

Tracked via

“shadow ai”