Researchers developed an attack technique named Adversarial Octopus that could perform a targeted attack on AI-based facial recognition systems. This attack shows that AI systems require much more attention at the security front, and such new attack methods will help raise awareness.