One operator uses AI to attack more than 100 companies
Forbes
Forbes
Dark Reading
Palo Alto Networks Unit 42
ExtraHop1 AFFECTED
The Register+12 AFFECTED
Tenable
CNN+11 AFFECTED
CNBC5 AFFECTED
CNN1 AFFECTED
DseWiki+41 AFFECTED
Wikimedia Foundation+12 AFFECTED
Wikipedia1 AFFECTED
Attacks run by AI agents rather than people, mapped as they are reported. Every entry links to the reporting it came from, so you can check it yourself. Free to use.
Seen an agentic attack, or movement that looks like one starting? Tell us. A person reads every report and checks it against a source before anything appears here, so nothing you send goes live automatically.
A person reads the reporting and writes the entry. There is no scraper, so nothing lands here because a headline matched a keyword. New entries go up as incidents are reported, usually the same day.
An incident covered by six outlets is one row, not six. Where several victims share one root cause the entries carry the same campaign, so the count of campaigns is shown next to the count of entries.
Dates, figures and attribution come from the source. Where something was not stated, the field is left empty rather than filled with a reasonable guess.
Attribution in this field is rarely settled. Where researchers stopped short of naming a state or a group, so does this page. Entries are summaries of someone else's reporting, not independent findings of our own, and the link is there so you can go and read the original. Where a named victim has disputed an account, the entry says so.
Kill Switch screens what your own AI says and what your agents do, and stops the ones that should not happen. Free to start, 100 screened messages, no card.