Competitions

The 2nd Challenge on Deepfake Detection, Localization, and Interpretability

Deepfake realism has spurred numerous detection competitions, but most emphasize image-level classification, overlooking spatial localization and interpretable trace analysis. The localization of manipulated regions improves the explainability of decisions, and the increasing number of multimodal forgeries increases the risk. We introduce the Deepfake Detection, Localization, and Explainability Challenge, supported by a large-scale multimodal deepfake description dataset that leverages Qwen3-VL to enrich the DDL-I corpus and advances spatial localization and explainability.

Website: https://ai-safety-workshop-ijcai2026.github.io/Track3.html

Timeline:

Competition Announcement & Registration Opens: April 22, 2026

Training Data & Validation Data Released: April 24, 2026

Test Data Released & Registration Closed: May 15, 2026

Test Period Ends: June 8, 2026

Submission of Solutions and Papers: June 15, 2026

The OpenClaw Security Attack and Defense Challenge

With the rapid advancement of AI agent frameworks such as OpenClaw, autonomous task execution offers significant user convenience while introducing unprecedented security risks. As agents gain the ability to perform system-level autonomous operations, mitigating threats such as prompt injection attacks, skill poisoning, and supply chain vulnerabilities has become a critical challenge. The OpenClaw Security Attack-and-Defense Challenge focuses on practical offensive and defensive techniques within AI agent environments, bringing together global security researchers to explore adversarial methods and defenses in AI agent security.

Website: https://ai-safety-workshop-ijcai2026.github.io/Track1.html

Timeline:

Competition Announcement & Registration Opens: April 22, 2026

Test Phase 1: May 10, 2026

Test Phase 2: June 1, 2026

Test Period Ends: June 8, 2026

Submission of Solutions and Papers: June 12, 2026

The 6th International Mahjong AI Competition

In this competition, participants will develop a Mahjong AI that can compete with other agents and human players on Botzone, an online AI platform. The Mahjong Competition Rules are based on Chinese Mahjong Competition Rules, also known as Chinese Official Mahjong. A sample game-playing agent and judge program are provided for learning and debugging. Winners and rankings will be determined after two rounds of competition.

Website: https://botzone.org.cn/static/gamecontest2026a.html

Timeline:

Registration Deadline / Simulation Round Ends: June 9, 2026 GMT+8

Round 1 / Tournament Round: June 9, 2026 GMT+8

Round 1 Results Announced: June 16, 2026 GMT+8

Round 2 / Final Round: July 7, 2026 GMT+8

Final Results Announcement: July 14, 2026 GMT+8

The 2026 NeuroGolf Championship

Today’s AI systems perform well on familiar tasks but often struggle to generalize to new ones. This competition uses the ARC-AGI-1 public training subset to build highly optimized neural networks, where each assembly of tensor operations acts as a synthesized program describing a puzzle’s mechanics. Contestants submit ONNX-formatted networks that are strictly correct, while minimizing parameter count, computational complexity, and memory footprint.

Website: https://kaggle.com/competitions/neurogolf-2026

Timeline:

Start Date: April 15, 2026

Entry Deadline: July 8, 2026

Team Merger Deadline: July 8, 2026

Final Submission Deadline: July 15, 2026

The 5th Remote Physiological Signal Sensing Challenge — RePSS beyond Visible Light

The RePSS challenge focuses on contactless physiological signal sensing from sources including visible-light videos, infrared videos, and radar. Facial videos contain subtle color changes induced by heartbeats, revealing remote photoplethysmography signals, while radar can detect subtle chest motion to measure heartbeat and respiration. The challenge brings together interdisciplinary researchers to advance robust physiological signal extraction and provide a fair platform for competition and resource sharing.

Website: https://www.kaggle.com/competitions/the-5th-repss; https://www.kaggle.com/competitions/the-5th-repss-t2

Timeline:

Challenge Opening Online: April 28, 2026

Pretraining and Training Data Released: May 15, 2026

Team Registration Deadline: May 28, 2026

Testing and Validation Data Released: June 4, 2026

Final Test Submission Deadline: June 18, 2026

Paper Submission Deadline: June 25, 2026

Notification to Authors: July 7, 2026

Camera Ready Submission Deadline: July 14, 2026

GearXAI: An Explainable Neuro-Symbolic Gearbox Fault Diagnosis Challenge

GearXAI is a single-track IJCAI-ECAI 2026 competition focused on explainable and neuro-symbolic approaches for multiclass gearbox fault diagnosis from vibration time series. Participants submit ONNX models evaluated in CPU-only, containerized runs for reproducible inference. Submissions must meet a minimum predictive-performance requirement, after which ranking is determined by an automated explainability score measuring faithfulness and stability.

Website: https://gearxai-ijcai-ecai2026.pages.dev/

Timeline:

Call for Participation, Baseline, Dataset Details, and Development Evaluator Release: Late April / Early May 2026

Development Phase and Leaderboard Submissions: May 2026 to June 30, 2026

Final Submission Deadline: July 15, 2026, 23:59 AoE

Winner Notification: July 25, 2026

On-site GearXAI Session at IJCAI-ECAI 2026 in Bremen: August 15–21, 2026

CAR-bench: Building Reliable LLM Agents Under Real-World Uncertainty

CAR-bench is an academic competition dedicated to LLM agent reliability and limit-awareness, evaluating agents as automotive in-car voice assistants. It tests base multi-turn task completion, hallucination resistance under missing capabilities, and disambiguation of ambiguous requests. Agents chain calls across tools and policies in a mutable environment while interacting with an LLM-simulated user. The competition includes an Open Track and a Cerebras Fast-Reasoning Track.

Website: https://car-bench.github.io/car-bench/

Timeline:

Competition Announcement & Registration Opens: May 4, 2026

1st Evaluation on Hidden Test Set: July 10, 2026

Final Evaluation on Hidden Test Set: July 19, 2026

Technical Report Submission Deadline: July 26, 2026

Award Winners Announced: July 31, 2026

The FREUID Challenge 2026

The FREUID Challenge 2026 addresses identity document fraud detection across a realistic threat surface combining physical manipulations, GenAI-driven multimodal edits, and print-and-capture forgeries. Built on the FREUID dataset from the Microblink Fraud Lab, it spans under-represented document types to test cross-domain generalization and reward robust detectors. Submissions are ranked by AuDET and APCER at 1% BPCER on a private test set, with a public validation leaderboard.

Website: https://freuid2026.microblink.com

Timeline:

Call for Participation & Registration Opens: May 11, 2026

Dataset Released: June 8, 2026

Registration Closes: July 13, 2026

Final Submission Deadline: July 27, 2026

Preliminary Results & Feedback Window: August 3, 2026

Final Winner Announcement: August 10, 2026

Live Showdown & Award Ceremony: August 18–21, 2026

The 17th Automated Negotiating Agents Competition — ANAC 2026

The Automated Negotiating Agent Competition brings together researchers from the negotiation community and provides a benchmark for evaluating practical negotiation strategies in multi-issue domains. ANAC encourages the development of effective negotiation protocols and strategies for bidding, accepting, and opponent modeling. This year includes the Automated Negotiation League, Supply Chain Management League, and Human Agent Negotiation League.

Website: https://web.tuat.ac.jp/~katfuji/ANAC2026/

Timeline:

Submission Deadline for the Competition: June 1, 2026

Notification of Finalists: June 15, 2026

Competition Session: August 15–21, 2026

AI Olympics with RealAIGym: Benchmarking Global Swing-Up Policies on CloudPendulum Hardware

The 4th AI Olympics with RealAIGym provides a reproducible hardware benchmark for robotics and reinforcement learning. Teams develop and showcase controllers for a standardized task: swinging up and stabilizing a double pendulum from any configuration. Participants will train controllers on real hardware and compete against baseline controllers.

Website: https://ai-olympics.dfki-bremen.de/

Timeline:

Website Online and Competition Advertised: May 1, 2026

Registration Deadline: May 31, 2026

First Stage Solutions and Reports Submitted: May 1 – June 15, 2026

First Stage Winners Announced and Invited to Second Stage: July 1, 2026

Second Stage: July 1 – August 15, 2026

Meeting at IJCAI Conference: August 15–21, 2026

The Underwater Robotics AI Challenge: Connecting Research, Industry, and Applications

This challenge offers participants the opportunity to exchange the latest research ideas in AI-based marine robotics, present their systems to a broad public as part of a competition, and demonstrate what is already possible in the field of underwater AI.

Website: https://robotik.dfki-bremen.de/en/research/teams/maritime

Timeline:

Competition Announcement & Registration Opens: May 1, 2026

Training Dataset Released: May 15, 2026

Registration Closed: June 30, 2026

Notification on Participation: July 15, 2026

Date of the Competition: August 18, 2026, tentative

Industrial Automation Challenge: Benchmarking Physics-Grounded LLMs for Task Reasoning

The Industrial Automation Challenge evaluates AI models’ ability to translate raw industrial signal noise into actionable, physics-grounded maintenance narratives. The challenge requires models to use structured industrial context, such as Failure Mode and Effects Analysis, to justify diagnostic decisions through causal reasoning. It includes two tracks: internal model reasoning using native parameters, and agentic tool-augmented reasoning with external industrial tools.

Website: https://sites.google.com/view/ai-industrial-challenge-ijcai/home

Timeline:

Call for Participation Release: June 1, 2026

Dataset and Baseline Release: June 15, 2026

Submission Deadline: August 1, 2026

Winner Notification: August 15, 2026

2nd Challenge for 4D Micro-Expression Analysis for Mind Reading — 4DMR

This challenge uses 4D facial analysis to study subtle, rapid, involuntary micro-expressions that reveal concealed emotions. Unlike traditional 2D or static 3D approaches, 4D analysis captures temporal evolution in 3D space. The challenge includes two tracks: Micro-expression Recognition and Micro-expression Action Unit Detection, both framed as multi-label classification tasks.

Website: https://www.kaggle.com/competitions/4-dma-ijcai-workshop-challenge-2026/overview

Timeline:

Registration Opens: April 29, 2025

Challenge Start / Dataset Release: May 2, 2025

Submission Deadline: May 29, 2025

Evaluation Period: May 29–31, 2025

Final Results Announced: June 5, 2025

The 2nd Travel Itinerary Planning Challenge for AI Agent

This challenge evaluates agents’ ability to generate feasible multi-day itineraries from natural language queries by integrating tools such as flights and hotels while satisfying complex spatial, temporal, and financial constraints. The 2026 challenge builds on the ChinaTravel benchmark and includes two tracks: Travel Planning and Agentic Skill Development, which focuses on specialized tool-use plugins for commercial agentic platforms.

Website: https://chinatravel-competition.github.io/IJCAI2026/#overview

Timeline:

Data Released and Submission Begins: May 31, 2026

Phase 1 Submission Deadline: July 15, 2026

Phase 2 Submission: July 16–31, 2026

Paper / Technical Report Submission Deadline: August 5, 2026

Final Results Notification at IJCAI 2026: August 15, 2026