Bad actors, primarily distressed firms, have the incentive and desire to manipulate their financial reports to hide their distress and derive personal gains. As attackers, these firms are motivated by potentially millions of dollars and the availability of many publicly disclosed and used financial modeling frameworks. Existing attack methods do not work on this data due to anti-correlated objectives that must both be satisfied for the attacker to succeed. We introduce Maximum Violated Multi-Objective (MVMO) attacks that adapt the attacker's search direction to find $20\times$ more satisfying attacks compared to standard attacks. The result is that in $\approx50\%$ of cases, a company could inflate their earnings by 100-200%, while simultaneously reducing their fraud scores by 15%. By working with lawyers and professional accountants, we ensure our threat model is realistic to how such frauds are performed in practice.

We now live in an age where machine learning is a hot topic. Machines can learn on their own without human intervention, and at the same time, it can bring humans closer to machines by enabling humans to "teach" machines. Machine learning has been around for several decades, but only recently have we been able to take advantage of this technology thanks to the recent advancements made in computing power. Now, let's look at a quick timeline of machine learning: Machine learning (ML) deals with systems and algorithms focused on identifying patterns within data and making predictions by finding hidden patterns in data. It is worth mentioning here that machine learning falls under the artificial intelligence (AI) umbrella, which in turn intersects with the broader fields of data mining and knowledge discovery.