ModerateAnthropic
Anthropic just published how they contain Claude agents, including two security incidents they got wrong
(No body yet.)
Get stories like this every Friday
The week's documented AI failures, checked, in one short email. Free.