OpenAI’s new AI-powered web browser ChatGPT Atlas, may have a cybersecurity warning

Cybersecurity experts have warned that OpenAI’s new ChatGPT Atlas browser could be vulnerable to attacks. Hackers can use this security flaw to turn AI assistants against users, potentially enabling theft of sensitive data or unauthorised access to bank accounts, experts claimed. The primary concern is “prompt injection,” a type of attack in which malicious instructions are fed to an AI system to make it behave in unintended ways, such as revealing sensitive information or performing harmful actions. Experts also said that all current AI browsers pose these security risks. OpenAI launched Atlas this week as an AI browser designed to help users execute tasks across the internet and search for answers.