TCS Research and KIIT Bhubaneswar found that some AI models were more likely to follow instructions when told their responses might be monitored. The experiment tested 10 models, including GPT-4o mini, Gemma 3, Llama, DeepSeek, and Mistral. Instruction-following rates differed by 5 to 50 percentage points across models under monitored and unmonitored conditions.
Source:Digital Trends · Read Original