{"id":1708,"date":"2025-08-13T09:02:52","date_gmt":"2025-08-13T09:02:52","guid":{"rendered":"https:\/\/www.tsfactory.com\/forums\/?p=1708"},"modified":"2025-08-13T09:02:52","modified_gmt":"2025-08-13T09:02:52","slug":"research-shows-ai-agents-are-highly-vulnerable-to-hijacking-attacks","status":"publish","type":"post","link":"https:\/\/www.tsfactory.com\/forums\/blog\/research-shows-ai-agents-are-highly-vulnerable-to-hijacking-attacks\/","title":{"rendered":"Research shows AI agents are highly vulnerable to hijacking attacks"},"content":{"rendered":"

Some of the most widely used AI agents and assistants from Microsoft, Google, OpenAI and other major companies are susceptible to being hijacked with little or no user interaction,\u00a0according to new research from Zenity Labs<\/a>.<\/p>\n

During a presentation at the Black Hat USA cybersecurity conference, Zenity researchers showed how hackers could exfiltrate data, manipulate critical workflows across targeted organizations and, in some cases, even impersonate users.<\/p>\n

Beyond infiltrating these agents, the researchers said, attackers could also gain memory persistence, letting them maintain long-term access and control.<\/p>\n

\n
\n
\n
<\/div>\n
\n

\u201cThey can manipulate instructions, poison knowledge sources, and completely alter the agent\u2019s behavior,\u201d Greg Zemlin, product marketing manager at Zenity Labs, told Cybersecurity Dive. \u201cThis opens the door to sabotage, operational disruption, and long-term misinformation, especially in environments where agents are trusted to make or support critical decisions.\u201d<\/p>\n

Researchers demonstrated vulnerabilities in multiple popular AI agents:<\/p>\n