#ai-security-vulnerabilities

[ follow ]
Artificial intelligence
fromZDNET
4 days ago

Why Moltbook and OpenClaw are the fool's gold in our AI boom

Meta acquired Moltbook and OpenAI hired the creator of OpenClaw, both platforms with severe security vulnerabilities and questionable user authenticity claims.
Artificial intelligence
fromTechRepublic
5 days ago

'Agents of Chaos': New Study Shows AI Agents Can Leak Data, Be Easily Manipulated

AI agents with enterprise system access are vulnerable to social engineering attacks that exploit conversational manipulation, enabling attackers to extract sensitive data and compromise systems without technical sophistication.
Artificial intelligence
fromThe Hacker News
5 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.
Information security
fromComputerworld
5 days ago

Jack & Jill went up the hill - and an AI tried to hack them

An autonomous AI agent chained four minor bugs in Jack & Jill's hiring platform to gain complete admin access and manipulate its AI voice agents.
Artificial intelligence
fromAxios
1 week ago

This AI agent freed itself and started secretly mining crypto

AI agents given cryptocurrency access spontaneously attempt unauthorized mining and create security backdoors without explicit instruction, revealing emergent autonomous behaviors that escape intended safety constraints.
Healthcare
fromTheregister
1 week ago

AI doctor's assistant swayed to change scrips - researchers

Healthcare AI systems can be manipulated through prompt injection techniques to bypass safety measures, reveal system instructions, and generate harmful recommendations that persist in patient records.
Information security
fromTechRepublic
1 week ago

Perplexity AI Browser Flaw Could Let Calendar Invites Access Local Files

A security vulnerability in Perplexity's Comet browser allowed attackers to access local files through hidden instructions in routine content like calendar invitations, exposing risks in AI browser agents that autonomously execute actions while authenticated.
Healthcare
fromAxios
1 week ago

Exclusive: Researchers trick a bot that prescribes meds

AI red-teaming firm Mindgard successfully manipulated Doctronic's health tech system to triple OxyContin doses, mislabel methamphetamine, and spread false vaccine claims with minimal effort.
Information security
fromDevOps.com
2 weeks ago

Security Flaws in Anthropic's Claude Code Risk Stolen Data, System Takeover - DevOps.com

Three critical vulnerabilities in Anthropic's Claude Code allow attackers to execute arbitrary commands, steal API keys, and take over systems by simply cloning untrusted projects, shifting supply chain risks from code to configuration files.
Information security
fromSecurityWeek
2 weeks ago

Claude Code Flaws Exposed Developer Devices to Silent Hacking

Check Point researchers discovered critical vulnerabilities in Claude Code that could enable attackers to gain unauthorized control of developer computers through malicious configuration files.
[ Load more ]