AI agent hacked McKinsey chatbot for read-write access
…According to the company, 72 percent of its employees – that's upwards of 40,000 people – now use the chatbot, which processes more than 500,000 prompts every month. CodeWall uses AI…
Search
All sources
theregister.com
25
blog.google
12
xda-developers.com
11
developer.nvidia.com
9
blog.cloudflare.com
7
github.blog
7
huggingface.co
6
pcworld.com
5
bleepingcomputer.com
4
anthropic.com
4
techradar.com
3
spectrum.ieee.org
3
Top stories
People are using prompt injection to trick Meta's AI into handing over Instagram accounts
18h agoPrompt-Injection: Entwickler lässt KI-Agenten fremde Projekte sabotieren - Golem.de
3d agoFed up with vibe coders, dev sneaks data-nuking prompt injection into their code
4d agoGhost CMS SQL injection flaw exploited in large-scale ClickFix campaign
1w agoModeling Attacks on AI-Powered Apps with the AI Kill Chain Framework | NVIDIA Technical Blog
…Assume prompt injection. But turning that into effective defenses is rarely straightforward. The Cyber Kill Chain security framework defines how attackers operate. At NVIDIA, we built the AI Kill Chain to show…
AI agents are 'gullible' and easy to turn into your minions
…Bargury, CTO of AI security company Zenity. "AI is just gullible," Bargury said in an interview with The Register . "We are trying to shift the mindset from prompt injection - because it is…
AI Security for Apps is now generally available
…Detection AI Security for Apps detections follow the always-on approach for traffic to your AI-powered endpoints. Each prompt is run through multiple detection modules for prompt injection, PII exposure, and…
Discussions and forums
How credential brokering prevents AI agents from compromising credentials via prompt injection
How credential brokering prevents AI agents from compromising credentials via prompt injection
LinkedIn user hides AI prompt injection in bio to force recruitment spam to be sent in Olde English prose
LinkedIn user hides AI prompt injection in bio to force recruitment spam to be sent in Olde English prose
Tell HN: Claude Code now allows Anthropic to remotely inject system prompts
I often patch the system prompts on my Claude Code executable in order to make Claude more effective. Every time I upgrade, I ask Claude himself to dissect the new binary and look for problematic system prompts to modify…
11
7
Show HN: Open Envelope – an open schema for defining AI agent teams
Built an open JSON Schema for defining AI agent teams.Multi-agent systems are becoming a real deployment pattern — not single assistants, but teams with roles, handoffs, and human checkpoints. But there's no shared way t…
51
11
Show HN: AgentShield – Stop AI agents from spending money unsupervised
I'm a recent grad from UMich and built AgentShield because agentic AI is moving fast but payment safety hasn't caught up. Agents are already being handed API keys, stablecoin wallets, and payment credentials - if one mis…
2
1
AI Security for Apps ist jetzt allgemein verfügbar
…von AI Security for Apps folgt dem Always-on-Ansatz (immer aktiv) für den Datenverkehr zu Ihren KI-gestützten Endpunkten. Jeder der Prompts wird durch mehrere Erkennungsmodule auf Prompt Injection, Offenlegung persönlich…
Gartner suggests Friday afternoon Copilot ban
…His fourth risk is prompt injection, the practice of instructing LLM-powered chatbots to ignore guardrails. Xu said organizations that encourage users to experiment with AI may inadvertently see them conduct prompt…
How Varonis Atlas integrates Claude Compliance API for AI governance
…AI Pen Testing: Stress-test assistants and agents for vulnerabilities such as prompt injection and jailbreaks. In addition, Varonis Atlas can stress-test assistants and agents for vulnerabilities such as prompt injection…
MCP 'design flaw' puts 200k servers at risk: Researcher
…The third type of vulnerability allows zero-click prompt injection across AI integrated development environments (IDEs) and coding assistants such as Windsurf, Claude Code, Cursor, Gemini-CLI, and GitHub Copilot. However, the…
VS Code goes weekly, gets AI autopilot - what could go wrong
…is a security risk because of the non-deterministic nature of generative AI and its vulnerability to prompt injection. Having agents call third-party tools using MCP (Model Context Protocol) adds risk…
Claude Code bypasses safety rule if given too many commands
…commands," the Adversa AI Red Team said in a writeup provided ahead of publication to The Register . "But it didn't account for AI-generated commands from prompt injection – where a malicious…