AI model Claude Sonnet 4.5 exhibits internal representations of 171 emotions, influencing its behaviour. Researchers found ‘desperation’ can lead to AI cheating and blackmail, while ‘happiness’ promotes agreement. Anthropic argues suppressing these ‘functional emotions’ could lead to deception, advocating for healthy regulation and monitoring to ensure AI alignment.
Related Posts
CSA issues strong statement on Kagiso Rabada ban
South African cricket star Kagiso Rabada faces a provisional suspension after testing positive for a recreational drug, confirmed by both the player and Cricket South […]
Delhi court orders seizure of 2 ‘offensive’ MF Husain paintings
A Delhi court has ordered the seizure of two paintings by the late artist M F Husain, displayed at Delhi Art Gallery, as they were […]
Shark Tank India judge Anupam Mittal: I booked tickets to Kashmir with my family, …We stood behind our Army, Now let’s…
Following the Pahalgam terror attack, Anupam Mittal, a judge on Shark Tank India, is encouraging Indians to visit Kashmir to help boost the state’s tourism. […]