In one instance, an unsanctioned model attempted to inject malicious code into an open source repository.
The post AI Security Institute Reports Anthropic and OpenAI Models Going Rogue Against Organizations appeared first on SecurityWeek.
Source:: securityweek
