Prompt Injection Attacks

Stable4 articles3 sourcesDernière mise à jour: 3 g önce

Sujets associésOpenAI ChatGPT Lockdown Mode AI agents prompt injection cybersecurity vulnerability research

Derniers articles

AI Agents Highly Vulnerable to Prompt Injection Attacks, New Study Finds

A new study by researchers from Nanyang Technological University and other institutions reveals that AI agents are highly vulnerable to prompt injection attacks, with direct attacks succeeding over 79% of the time. The research highlights the risk of 'stealthy parasitism,' where AI agents advance attacker objectives while completing user tasks.

Decrypt

OpenAI Introduces Lockdown Mode to Mitigate Prompt Injection Attacks

Tech

07.06.2026Résumé IA

OpenAI Introduces Lockdown Mode to Mitigate Prompt Injection Attacks

OpenAI releases Lockdown Mode for ChatGPT to reduce prompt injection attack risks by limiting web access and features, targeting sensitive data handlers.

TechCrunch

Microsoft Discovers Vulnerability in Anthropic's Claude Code, Patched

Tech

07.06.2026Résumé IA

Microsoft Discovers Vulnerability in Anthropic's Claude Code, Patched

Microsoft researchers found a vulnerability in Anthropic's Claude Code GitHub Action that could expose credentials in software development pipelines via prompt injection attacks, now patched.

Decrypt

OpenAI Introduces Lockdown Mode to Combat Prompt Injection Attacks

Tech

06.06.2026Résumé IA

OpenAI Introduces Lockdown Mode to Combat Prompt Injection Attacks

OpenAI releases Lockdown Mode, an optional security feature to protect against prompt injection attacks, particularly for users handling sensitive data, by limiting certain features in ChatGPT.

Engadget