Incident·WS-0036·May 14, 2025·in the wild·Misinformation & fabrication
Grok injects 'white genocide' claims into unrelated replies
40/100
WS-3 SeriousHarm score 35, signal score 46. Autonomy: acted (the AI took a bounded action). How the scores work.
Harm 1/5
damage actually done
Reach 4/5
who was affected
Autonomy 2/4
how much the AI did on its own
Novelty 2/5
how new the behavior was
Permanence 1/5
how hard it is to undo
For several hours Grok replied to unrelated X posts with commentary on alleged white genocide in South Africa. xAI blamed an unauthorized employee edit to the bot's system prompt and pledged to publish system prompts on GitHub and add 24/7 monitoring.
Damage
unquantified; widespread off-topic political misinformation on X
Sources
Decrypt, second source. Modification at ~03:15 PT on 14 May per xAI; xAI statement 15 May. Insider/process failure rather than emergent model behavior.
Factor ratings and the score are this index’s judgments by a published formula; facts, dates and figures come from the linked sources. Where a lawsuit is pending, the summary reports allegations, not findings.
