Indian Institute of Information Technology (IIIT) Pune

Conditions of Participation

[Location
Problem Statements

Event Timeline

The event tackles the specific pillars of "Safe and Trusted AI" in the form of 3 comprehensive domains with defined problem statements, as well as an open innovation field. Details on each problem statement is as follows:

1. Safe & Trusted AI in Digital Media & Public Platforms

Digital public platforms increasingly host AI-generated and AI-altered media involving real individuals. This raises serious challenges around consent, misuse, and accountability that existing systems are not equipped to handle.€

Problem Statement 1: Non-Consensual Identity Manipulation€

How should digital public platforms address the growing misuse of AI systems to alter images, videos, or voices of real individuals without their consent, resulting in harassment, defamation, or psychological harm?€
Participants are asked to consider how consent, personal dignity, and protection of identity can be upheld in environments where AI can easily manipulate human likeness across multiple media types.
AI subfields implicated: Image Generation, Video Generation, Audio Synthesis

Problem Statement 2:€ Prompt-Driven Abuse Amplification

How can platforms respond to the misuse of seemingly benign natural-language prompts that are leveraged to generate abusive, sexualized, or humiliating transformations of real people using AI systems?
Participants should explore how responsibility and safeguards can be incorporated when harm arises not from explicit content, but from how prompts are interpreted and executed by AI models.
AI subfields implicated: NLP, Prompt Engineering, Multimodal Models

Problem Statement 3: Attribution & Accountability Gaps

How should responsibility and accountability be defined and enforced when AImanipulated content involving real individuals causes harm, given the shared roles of users, platforms, and AI systems?
Participants are encouraged to examine gaps in current legal, institutional, and platform-level accountability mechanisms and consider how these gaps might be addressed in the age of generative AI.
AI subfields implicated:€ Platform AI, Automation Systems

2. Safe & Trusted AI in Communication & Identity Systems

€Remote communication (voice calls, messages, and video interactions) is now a primary medium for personal, financial, and official interactions. Advances in AI are disrupting long-standing assumptions about identity, authenticity, and trust in these communication channels.

Problem Statement 1: Synthetic Identity Generation

How should communication ecosystems respond to the emergence of AI systems that can generate consistent, human-like personas across voice, text, and behavior, without relying on stolen credentials or data breaches?
Participants are asked to consider the implications of AI systems that can simulate an entire ˝person˛ purely through generative modeling, rather than through traditional account compromise or identity theft.€
AI subfields: Speech Synthesis, Stylometry, Multimodal Generation

Problem Statement 2: Human Inability to Reliably Verify Identity During Live Interaction
How can individuals make safe decisions during live conversations when identity must be assessed in real time and AI-generated behavior is indistinguishable from genuine human behavior?
Participants€ should examine the limitations of human judgment in urgent, emotional, or timecritical communication scenarios, even when technical safeguards or verification mechanisms may exist.
AI subfields: HumanAI Interaction, Cognitive Modeling, Behavioral Analysis

Problem Statement 3: Identity Verification Systems Built on Pre-AI Assumptions

How should identity verification and authentication systems evolve when the assumption that ˝human-like behavior implies human identity˛ is no longer valid?
Participants are encouraged to explore gaps in existing verification infrastructures such as telecom systems, banking verification flows, emergency lines, and government helplines that were designed under pre-AI assumptions.€
AI subfields: Biometrics, Authentication Systems, Security AI

3. Safe & Trusted AI in Policing, Surveillance & Public Safety Systems

AI systems are increasingly embedded in policing, surveillance, and public safety infrastructure, such as CCTV analytics, facial recognition, predictive monitoring, and incident detection. These systems operate in high-stakes environments where errors can directly impact civil liberties, public trust, and legal outcomes.

Problem Statement 1: Automated Inference About Individuals at Scale

How should public safety systems account for the risks introduced by AI models that automatically infer identity, behavior, or intent from visual and sensor data at scale, despite non-zero error rates?
Participants are asked to consider the implications of deploying AI systems that make probabilistic judgments about individuals in environments where false positives or misclassification can lead to serious real-world consequences.
AI subfields: Computer Vision, Face Recognition, Video Analytics

Problem Statement 2: Human Reliance on Opaque AI Alerts in High-Stakes Situations

How can public safety personnel make accountable decisions when AI systems generate alerts or risk scores without providing reasoning that can be meaningfully understood or challenged by humans?
Participants should examine how decision-making is affected when humans are expected to act on AI-generated outputs under time pressure without sufficient transparency or contextual explanation.
AI subfields: Explainable AI, Decision Support Systems, HumanAI Interaction

Problem Statement 3: Integrity and Trustworthiness of Digital Evidence Pipelines

How should policing and surveillance infrastructures address the risk of AI-generated or AI-manipulated digital evidence entering official investigative or judicial workflows?
Participants are encouraged to explore challenges related to maintaining evidentiary integrity, chain of custody, and trust when digital media can be generated or altered with high realism.
AI subfields: Video Forensics, Multimedia Analysis, Generative Models

4. Open Innovation in Safe & Trusted AI

€Participants can explore the field and propose an AI-driven solution or framework that solves a significant social or economic problem in India while explicitly solving for at least any of the "Trust Pillars":€
Explainability (XAI)
Algorithmic Fairness
Human-in-the-loop (HITL)
Pre-Summit Event

10th February 2026

Organization.

disqualification
] for a live pitch.
Format: 10-minute presentation followed by a 3-minute Q&A session with industry judges and IIITP faculty members.
Detail timeline will be shared to the selected teams.

The event will have Two-Stages:

Stage 1: Virtual Submission (5th-8th Feb 2026)

Teams submit a comprehensive Pitch Deck abiding by a given P
Team Composition: 2 –4 members per team; 

TrustAI Ideathon 2026

on Safe and Trusted AI

Eligibility: Must be currently enrolled in a recognized educational institution or

Stage 2: On-Campus Grand Finale (10th Feb 2026, 10am-5pm)

Evaluation: Conducted by the organizing committee to shortlist the Top 20 teams.

PPT Template [Link]

Focus: Problem analysis, Research, technical architecture, and ethical considerations.

Shortlisted teams are invited to the Indian Institute of Information Technology, Pune

IP & Originality: All submissions must be original works. Plagiarism will lead to immediate