You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
We are requesting funding for conference travel (iAIMS 2026/IFIP) and open-access publications (APCs) to communicate our 8-paper technical research pipeline. The research provides usable Behaviour Reconstruction Models (BRMs) and Responsibility Attribution Frameworks (RAFs) which enable the mapping of log telemetry, adversarial runtime exploits, and the closing of the accountability gap in autonomous AI agent cyber incidents.
Our goal is to address the “responsibility attribution gap” in autonomous AI agent cyber incidents that exists before agentic deployments reach a scale that surpasses the pace of international safety governance. In the case of a security incident resulting in model drift, reward hacking or indirect prompt injection by an autonomous agent, the conventional forensic logs cannot distinguish between an intentional attack on the model and an anomaly in the model.
In this regard, we researched and proposed a dual-layer technical solution: a Behaviour Reconstruction Model (BRM) to reconstruct the internal chain of thought of an agent post-incident and a socio-technical Responsibility Attribution Framework (RAF) to map the legal responsibility among developers, deployers and other stakeholders.
We plan to communicate this research pipeline in the next 12 months as follows:
Present our accepted paper on adversarial memory laundering at iAIMS 2026 in Barcelona, Spain.
Deliver a practical log inspection guide to thousands of active incident response examiners at the Magnet Virtual Summit.
Present our benchmark paper on state-graph checkpointing at the IFIP WG 11.9 International Conference on Digital Forensics in New Delhi.
Publish 5 more papers currently under journal review (Discover Artificial Intelligence and Artificial Intelligence and Law) in Open Access formats, to maximise reach and visibility.
We plan to use the requested funding ($7,710) for conference travel, event registration, and journal publication fees (APCs).
The exact cost allocation is:
Conference Registration ($800) for iAIMS2026 + ($920) for IFIP WG 11.9.
IEEE and Computer Society membership ($160)
Round-trip to Spain ($1,100)
C-Type Schengen visa with required travel health insurance ($180).
Accommodation ($1,100)
Local transport & meals ($450).
Journal Publication Pool ($3,000)
This project is a direct research collaboration between:
Ahmad Muhammad Tahir (Myself): a postgraduate student pursuing an MSc in Digital and Cyber Forensics at the Department of Forensic Science at Vivekananda Global University. Alongside my postgraduate research, I serve as a Senior Technologist in the Department of Computer Science at Aliko Dangote University of Science and Technology, where I actively head the department's Embedded and Networking Laboratory.
Ms Garima Joon is an Assistant Professor within the Department of Forensic Science at Vivekananda Global University. She is a veteran cybersecurity researcher specialising in cloud evidence acquisition and mobile forensics tooling, with a proven track record of multiple publications and patents in the field.
We have demonstrated our execution capability on this specific agentic security framework through our completed 12-month pipeline:
Accepted: Our paper on RAG runtime injections has been accepted for oral presentation at iAIMS 2026 in Barcelona, Spain.
Active Review: We have papers under active review at Discover Artificial Intelligence (awaiting editor decision), Information Security Journal: A Global Perspective, Cureus Journal of Medical Science, and IFIP WG 11.9 / AISSG 2026.
We have submitted operational DFIR logs to the Magnet Forensics user summits.
Our primary risk of failure is a dissemination bottleneck caused by lack of liquidity.
Cause of Failure: If funding is insufficient, I will not be able to attend and present the papers at iAIMS2026 and IFIP WG 11.9. The journal publications will have to be closed under subscription-based paywalls.
Outcome of Failure: Despite the need for open-source logging primitives to safely control deployed agent behaviours, the research will stay confined within academia, away from policymakers and industry experts
$0. This research has been entirely self-funded and completed independently without active grants, corporate sponsorships, or university institutional R&D pools.
There are no bids on this project.