By: Madison Bruner

Every young student knows the dangers of cheating, plagiarizing, and deceiving through their academic work. But what happens when these students become high-power pharmaceutical scientists and mislead the research world through scientific fraud? The stakes are higher, and the consequences are much more severe. Recently, an action as small as manipulating a western blot image has snowballed into a massive instance of scientific fraud, with millions of dollars in damages and potential criminal convictions. Although the information in this post highlights scientific fraud, this instance of fraud is an anomaly within an otherwise upstanding scientific community.

What is a Western Blot?

The western blot is a scientific research technique used in molecular biology, immunogenetics, and pharmacology research. The method uses electricity to drive protein-rich tissue samples through a gel that separates the molecules by size, allowing for protein identification. Distinct proteins, tagged and illuminated by fluorescent antibodies, appear as stacked bands.  This technique has been used in pioneering research on tumor suppressor proteins, HIV proteins, and Alzheimer’s disease.

The Problem of Fraudulent Western Blots

With the expectation to consistently publish papers and engineer patentable research methods, scientists are under immense pressure to produce groundbreaking work and do so frequently.

Scientists have been exposed for falsifying data, including western blot images. Researchers have been found copying and pasting a protein sample from one well into another. Researchers have also been caught digitally altering western blot images by rotating, flipping, or stretching the image, reflecting results that are ingenuine.

Scientists may also use deepfakes, AI-created synthetic media, to create artificial western blots that prove difficult to detect. Scientists can do so via Generative Adversarial Networks—a type of machine learning network—that uses a database of images to create unique images distinct from existing western blot images. Even AI tools that are designed to detect fraudulent blots may struggle to identify these fraudulent images, especially if they are low resolution. However, the AI may flag the issue for potential misconduct, triggering further investigation.

Legal Consequences of Scientific Fraud: A Case Study

Sylvian Lesné of the University of Minnesota, Twin Cities, published a study in Nature in 2006 presenting what has become a dominant hypothesis of the cause of Alzheimer’s disease. The theory holds that amyloid beta clumps in brain tissue are the primary cause of the illness known to have no cure, which currently affects 7 million Americans. National Institute of Health (“NIH”) funding for research related to the amyloid beta hypothesis went from nearly zero to $287 million in 2021.

*Scientists raised concerns of Lesné’s western blot images included in a Nature publication researching amyloid-beta on PubPeer, a non-profit foundation dedicated to post-publication peer review.

Matthew Schrag, a fellow neuroscientist, spotted red flags in Lesné’s work, which prompted an investigation by Science. The investigation found issues with over 70 images included in Lesné’s papers, some with evidence of “shockingly blatant” image tampering. Some experts suspect Lesné’s studies have “misdirected Alzheimer’s research for 16 years.” Schrag also publicly criticized the FDA approval of an anti-amyloid beta pharmaceutical drug, Aduhelm. 

With Schrag’s assistance, attorneys petitioned the FDA to halt the clinical program for Simufilam, a similar anti-amyloid beta drug developed by Cassava Sciences. In June 2024, Hoau-Yan Wang, an Alzheimer’s researcher working for Cassava, was criminally indicted for allegedly defrauding the NIH of approximately $16 million in grants related to Simufilam research. The indictment alleged Wang manipulated western blots through his research. This matter is set for trial on September 22, 2025. In September 2024, Cassava settled for $40 million with the Securities and Exchange Commission (SEC) for misleading investors about earlier clinical trial results for Simufilam. 

AI and Fraud Detection

Given AI’s pervasive influence, it is no surprise that AI tools are assisting the battle against scientific misconduct. Previously, western blot fraud-spotting was conducted through visual observation, such as the massive review of 20,621 scientific papers from 1995 to 2014 by Elizabeth Bik, a science integrity consultant. Bik found that one in every twenty-five articles with western blot images contained “problematic” figures with features suggesting deliberate manipulation.

New AI tools such as Proofig and ImageTwin are utilized by journals to streamline the process of spotting fake western blot images. These tools use databases such as PubMed’s Source database to compare submitted images to existing images in publications. If flagged, scientists may have a chance to fix the error and resubmit their paper to the relevant journal, but sometimes flagging includes a complete retraction of the submission.

The ethical implications of AI use for fraud-spotting are serious. AI developers have an ethical duty to avoid biases and discrimination in AI outcomes and datasets, ensure scientists’ data is protected and anonymized, and regularly audit the tool with human oversight in order to catch and mitigate any potential harm. Journals and research institutions should remain transparent about the use of these tools.

Legal Considerations of Scientific Fraud for Scientists and Employers

In addition to criminal fraud charges and SEC enforcement actions, the rise of AI in detecting fraudulent scientific data brings profound legal implications for both employers and employees.

Further Regulatory Consequences

Federal regulations require that Institutions funded by Public Health Services, such as the National Institute of Health, report any threshold findings of misconduct. Institutions must have written policies and procedures to address allegations of research misconduct. The Office of Research Integrity (“ORI”) oversees compliance, reviews institutional findings, and may conduct its own investigation. The ORI can propose administrative actions to be taken against the researcher, including debarment, suspension, and supervision requirements. The ORI may also publish findings of research misconduct. Employees accused of fraud may contest ORI findings through a hearing before a Department of Health and Human Services Departmental Appeals Board Administrative Law Judge.

Civil Lawsuits

Uncovering scientific fraud brings negative publicity to both the associated institution or corporation and the researcher involved and opens the possibility that an employer may be sued for defamation, wrongful termination, product liability, negligence, or investor protection. Employers may update their employee policies or contractual agreements to include AI screening of article submissions to combat the recent rise in scientific fraud and reduce legal risks.

The advancement of AI in detecting fraudulent western blots and other scientific techniques not only fortifies scientific integrity but also introduces a new era of legal accountability. Institutions, corporations, and researchers are compelled to adhere to higher standards of transparency and ethical conduct in this continuously altering landscape.