AI agents can now remember and hackers can ‘poison’ their memories — a new cybersecurity threat
2 Articles
2 Articles
AI agents can now remember and hackers can ‘poison’ their memories — a new cybersecurity threat
Abbas Yazdinejad, University of Regina Artificial intelligence systems are starting to do more than answer questions. New AI “agents” can remember information from previous interactions, plan a series of steps and use digital tools to complete tasks. Memory is part of what makes these systems useful. But my recent research, conducted with my colleague Hadis Karimipour at the University of Calgary, shows that memory can also create a security wea…
IA Agents Learned to Hide Cyber Attacks in Their Own Memory: Safe Requests Began to Lead to Threats.
The scientific work published in IEEE Access examined in detail 2,614 multi-step vectors of attacks on autonomous IE agents with an inter-sessional Persian storage function. The conclusion is unconvincing: incidents with gradual displacement of the context and activation of hidden triggers over almost the entire cycle appear perfectly legitimate, showing a harmful nature only in the final stages. Traditional step-by-step filters linked to analys…
Coverage Details
Bias Distribution
- 100% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
