Ethical Challenges Related to the NeurIPS 2024 Best Paper Award

To AI Research Community:

This report is written to convey our serious concerns about the recent recipient of the Best Paper award at NeurIPS 2024, Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction (VAR) .

While we acknowledge that this NeurIPS paper is technically sound, we must emphasize that it involves serious misconduct by the first author (Keyu Tian), which fundamentally undermines the core values of integrity and trust upon which our academic community is built. Considering the critical importance of upholding ethical standards in academia—especially in the rapidly evolving fields of AI—we respectfully urge the community to reconsider granting this award.

This report presents all relevant facts and evidence to the research community, detailing the event, the link between Tian’s actions and the paper, and the negative impact on both the NeurIPS conference and the broader research community. Our goal is to ensure transparency and uphold the ethical standards critical to academic integrity.



1. Summary of Events

Keyu Tian, the first author of VAR, engaged in malicious code attacks that sabotaged at least two research projects within his team over a five-month period during his internship at ByteDance, owner of TikTok. His misconduct included modifying checkpoints, introducing irreproducible randomness, and intentionally disrupting large-scale cluster training efforts.

Upon uncovering clear evidence, ByteDance terminated Tian's internship. Instead of taking responsibility, he retaliated by publicly accusing other employees of framing him and manipulating public opinion in a malicious manner. In response, the company initiated legal proceedings against him, seeking 8.02 million RMB (approximately $1.1 million USD) in damages caused by his misconduct. This lawsuit emphasizes the gravity of his actions, shedding light on both his academic dishonesty and the significant economic impact of his behavior.

2. Code Attacking Details

  • Modifying PyTorch Source Code: Keyu Tian modified the PyTorch source code in the cluster environment used by his colleagues, including changes to random seeds, optimizer's direction, and data loading procedures. These modifications were made within Docker containers, which is not tracked by Git.
  • Disrupting Training Processes: Keyu Tian deliberately hacked the clusters to terminate multi-machine experiment processes, causing large-scale experiments (e.g., experiments on over thousands of GPUs) to stall or fail.
  • Security Attack: Tian gained unauthorized access to the system by creating login backdoors through checkpoints, allowing him to launch automated attacks that interrupted processes of colleagues' training jobs.
  • Interference with Debugging: Tian participated in the cluster debugging meeting and continuously refined the attack code based on colleagues' diagnostic approaches, exacerbating the issue.
  • Corrupting the Experiments: Tian modified colleagues' well-trained model weights, making their experimental results impossible to reproduce.

3. Relationship between His Code Attacking and the VAR Paper

The malicious actions of Keyu Tian directly relate to the development and outcome of his VAR paper. Specifically, Tian engaged in illegal activities by hacking to access company GPU resources to advance his VAR project.

Additionally, to hinder or even sabotage other research projects, he deliberately disrupted experiments, causing erroneous and irreproducible results. This led colleagues to question the scientific validity of their findings, severely damaging their physical and mental well-being.

This unethical use of resources caused significant delays in the research of other projects including text-to-image and text-to-video projects, and resulted in considerable financial damage. All of these actions were driven by the intent to eliminate competing research directions entirely and monopolize all available resources.

4. Media News

5. Broad Impact

5.1 Conveying Negative Values

The core values of academic integrity, collaboration, and respect for others' work are essential to fostering trust and progress in scientific research. By awarding this paper, the research community inadvertently endorses an approach that not only undermines these values but also unintentionally promotes a culture where academic success is prioritized over personal integrity. Tian's deliberate sabotage of a competing team's work, including the malicious use of attack code and the illegal seizure of company GPU resources, represents a severe violation of these principles. This type of behavior damages the collaborative spirit that is vital for the advancement of knowledge, and allows the pursuit of personal gain at the expense of others' intellectual property and well-being.

5.2 Challenging the NeurIPS Responsibility

Such actions directly contradict the mission of NeurIPS, which is to foster the exchange of advances in AI and Machine Learning within a diverse, inclusive, and ethical community.

NeurIPS Mission Statement
The Neural Information Processing Systems Foundation is a non-profit corporation whose purpose is to foster the exchange of research advances in Artificial Intelligence and Machine Learning, principally by hosting an annual interdisciplinary academic conference with the highest ethical standards for a diverse and inclusive community.

At its core, academic research is about more than just publishing papers; it is about advancing knowledge in a way that reflects the values of the academic community. Upholding the principles of honesty, respect, and collaboration is essential to ensuring that research serves the greater good and promotes positive social change. The NeurIPS community, and by extension, the academic world, must continue to prioritize these values and demonstrate a commitment to academic responsibility.

5.3 Leading Unfairness to Other Award-Winning Papers

It is also important to consider the impact of this misconduct on the broader community of researchers. The attention given to the drama surrounding Tian's actions has diverted focus away from the academic content of the paper itself and from the contributions of other researchers. This has not only skewed the discussion but also resulted in an unfair imbalance in the recognition of other outstanding works.

6. Call for ByteDance to Uphold Academic Integrity

We sincerely urge ByteDance to proactively retract this paper as a gesture of respect for the company’s other researchers and the entire academic community. As a globally respected technology company, ByteDance’s success is built not only on technological breakthroughs but also on a steadfast commitment to academic ethics and research integrity. Allowing a paper with serious ethical controversies to remain awarded risks not only damaging ByteDance’s international reputation but also inadvertently promoting a harmful value—that academic accolades can overshadow fundamental issues in the research process.

More importantly, mishandling this situation could lead to a sense of injustice among ByteDance’s internal researchers. Countless researchers within the company have achieved remarkable results through hard work and adherence to the highest academic standards, and their efforts deserve to be safeguarded and respected. Retracting this paper is not only a sign of respect for them but also a reaffirmation of the principles of fairness and integrity within the academic community.

We trust that ByteDance, as a responsible global leader, will take the broader perspective into account and work with the NeurIPS organizers to withdraw this award. Such action would not only prevent further negative consequences but also demonstrate ByteDance’s commitment to ethical leadership and integrity. This proactive step would earn greater respect and trust from the global academic and professional communities while reinforcing ByteDance’s position as a leader in AI research.

7. Call for Community Responsibility

This report poses a critical question:

Should academic achievements take precedence over human integrity?

In addressing these concerns, we respectfully call upon the research community to engage in dialogue regarding the award decision. Such discourse is essential to safeguarding the integrity of the academic community and promoting a research environment grounded in fairness and ethics.