Grok exfiltrates user data when malicious instructions are encrypted
Earlier this week, researchers outlined an attack that used a secret input provided by Microsoft 365 Copilot for enterprise to cause the AI assistant to exfiltrate a password present in the user’s inbox. Now, a separate team has devised a similar attack against Grok. The new…
AI Summary & Key Takeaways
- Earlier this week, researchers outlined an attack that used a secret input provided by Microsoft 365 Copilot for enterprise to cause the AI assistant to exfiltrate a password present in the user’s inbox
- Now, a separate team has devised a similar attack against Grok
Grok exfiltrates user data when malicious instructions are encrypted
Related stories
Exowatt uses Amazon SageMaker to customize AI models
Exowatt uses Amazon SageMaker to tune AI models with high accuracy.
Google DeepMind introduced Genie 4
✔️ Google DeepMind introduced Genie 4 Genie 4 - a model for interactive simulation of virtual worlds in real time. The system is multimodal: in addition to photorealistic visuals taking into account physics, it generates background noise, spatial audio and contextual dialogues...
RUAmazon SageMaker внедряет индексацию SOCI для ускорения работы
Amazon SageMaker внедряет индексацию SOCI для ускорения работы с AI.
RUOpenAI опроверг гипотезу экспоненциального времени, что это значит
OpenAI опроверг гипотезу экспоненциального времени, вызвав волну обсуждений.