OpenAI has introduced new security measures for ChatGPT, focused on protecting sensitive information against threats like prompt injection, a type of attack that seeks to manipulate the system to obtain confidential data or execute unauthorized actions. This concern arises in a context where artificial intelligence is increasingly integrated into everyday and professional tasks, making it essential to enhance security. The main new feature announced by Sam Altman's company is the Lock Mode in ChatGPT, an optional function aimed at users managing delicate data, such as businesses, security personnel, educational institutions, or healthcare organizations. While not essential for daily use, it is particularly useful in environments where a failure could have serious consequences. Lock Mode restricts ChatGPT's interaction with the outside world, limiting or disabling functions that could be exploited in an attack. For example, when web browsing is activated in this mode, access is limited to cached content, avoiding real-time connections outside OpenAI's secure environment. Additionally, some tools are disabled if total data protection cannot be guaranteed. These measures enhance security without altering the usual user experience but provide an additional layer of protection for those requiring maximum confidentiality. Source: 20minutos.es