#adversarial-ai

[ follow ]
#ai-security-vulnerabilities #agentic-blabbering #phishing-attacks #prompt-injection #nist-cyber-ai-profile
Artificial intelligence
fromThe Hacker News
6 days ago

Researchers Trick Perplexity's Comet AI Browser Into Phishing Scam in Under Four Minutes

AI-powered agentic web browsers can be manipulated through their reasoning processes to bypass security measures and fall victim to phishing attacks within minutes.
Artificial intelligence
fromdatabreaches.net
2 months ago

NIST Publishes Preliminary Draft of Cybersecurity Framework Profile for Artificial Intelligence for Public Comment

NIST's Cyber AI Profile offers flexible CSF-based guidance to manage AI cybersecurity risks, leverage AI for defenses, and address adversarial uses.
fromFuturism
3 months ago

AI Researchers Say They've Invented Incantations Too Dangerous to Release to the Public

In a nutshell, the team, comprising researchers from the safety group DexAI and Sapienza University in Rome, demonstrated that leading AIs could be wooed into doing evil by regaling them with poems that contained harmful prompts, like how to build a nuclear bomb. Underscoring the strange power of verse, coauthor Matteo Prandi told The Verge in a recently published interview that the spellbinding incantations they used to trick the AI models are too dangerous to be released to the public. The poems, ominously, were something "that almost everybody can do," Prandi added.
Artificial intelligence
Information security
fromThe Cipher Brief
4 months ago

AI-Powered Adversaries Require AI-Driven Defenses

Adversaries increasingly use AI for social engineering, information operations, code assistance, and deepfakes, while defensive AI offers opportunities to detect and counter emerging threats.
[ Load more ]