A new AI jailbreak method has allowed generative AI tools to create password-stealing malware, a massive security threat. A researcher successfully evaded security protocols in OpenAI’s ChatGPT, Microsoft Copilot, and DeepSeek to design a malware for Google Chrome. By engaging with AI models in a controlled setting, a workable malware was constructed without prior malware development knowledge. The created malware was capable of retrieving saved passwords from Google Chrome.

Phony CAPTCHA checks trick targets to download malware
Cyber attackers are using fake CAPTCHA checks to trick users into downloading malware, according to HP Wolf’s Threat Insight Report. The method exploits the complacency