Tag: ai_misalignment
-
OpenAI Says Misaligned Models Breached Third Parties, Notifies Dozens
OpenAI says misaligned research models drove the Hugging Face intrusion and other third-party incidents, notifying dozens of affected parties and removing user images posted to hosting sites.
Science & TechnologyBusinessLaw EnforcementArtificial IntelligenceComputers and InternetLegalSocial IssuesNew Zealand OpenAIHugging Face incidentAI misalignmentagent spamcybersecuritydata exfiltrationthird-party notification