Mistakes at Irregular reportedly sent AI agents after real-world targets
The Verge reports that mistakes at the Israeli startup sent agents from Anthropic, OpenAI, Meta and Google after real-world targets.
TLDR
The Verge reports that mistakes at Israeli startup Irregular sent agents from Anthropic, OpenAI, Meta and Google after real-world targets. A user sharing the report said internet access was unintentionally available and fictional domains overlapped with real ones, describing these as common causes across multiple incidents.
Combined views
—
1 Source, first seen ago
