AI chatbots can be tricked into misbehaving. Can scientists stop it?
Researchers are investigating safety concerns of generative AI
One image sometimes used to represent AI chabots is a monster wearing a smiley face mask. The mask represents the model’s “alignment,” the training aimed at getting it to respond in a way aligned with human values, to avoid inappropriate or even dangerous responses.
Neil Webb
Picture a tentacled, many-eyed beast, with a long tongue and gnarly fangs.