When prompts were presented in poetic rather than prose form, attack success rates increased from 8% to 43%, on average — a fivefold increase.
The original article found on darkreading Read More
Researchers Use Poetry to Jailbreak AI Models
Related Posts
Researchers Expose New Polymorphic Attack That Clones Browser Extensions to Steal Credentials
Cybersecurity researchers have demonstrated a novel technique that allows a malicious web browser extension to impersonate any installed add-on. “The polymorphic extensions create a pixel […]
Hackers Hiding NFC Carders Behind Apple Pay and Google Wallet
Cybercriminals have devised sophisticated methods to exploit Near Field Communication (NFC) technology via popular mobile payment platforms. These attackers are now leveraging Apple Pay and […]
Microsoft Backs Massive AI Push in UAE, Raising Security Concerns
In partnership with Emirates tech company G42, Microsoft is building the first stage of a 5-gigawatt US-UAE AI campus using Nvidia GPUs. The original article […]