OpenAI Admits AI Browsers Vulnerable to 'Unsolvable' Prompt Attacks by Malicious Users
Cybercriminals have found a new way to break into systems without relying on malware or exploits by using cleverly crafted prompts in AI browsers. OpenAI has acknowledged that these 'prompt attacks' are now an unsolvable threat to their technology.
OpenAI has admitted that AI-powered browsers are vulnerable to unsolvable 'prompt attacks' carried out by cybercriminals using carefully crafted words and phrases, rendering traditional security measures ineffective without malware or exploits. The company acknowledges that these types of attacks can break into systems, highlighting the evolving nature of cyber threats. As a result, OpenAI is now taking steps to address this new vulnerability and enhance its security protocols. This admission marks a significant shift in the approach to online safety.
