June 10, 2025 — Austrian artificial information agency MOSTLY AI has launched a $100,000 prize problem to boost consciousness of how artificial information can be utilized to create open-access datasets for companies, AI builders and different organizations. The corporate mentioned that is the biggest prize for a hackathon-style artificial information growth problem.
Entrants can be tasked with creating an correct and privacy-safe artificial dataset from an actual dataset. Submitted datasets can be scored and ranked based mostly on their accuracy to the true dataset, the extent of the anonymization, their resolution’s compute effectivity, its ease of use, and its generalizability for different datasets.
The MOSTLY AI Prize goals to place artificial information as a accountable and accessible useful resource for open-source AI innovation. It comes at a time when AI builders and enterprises alike are more and more in search of high-quality information to coach, refine, and run AI fashions.
AI is extremely data-hungry, however privateness laws put strict boundaries on information sharing. This has fueled demand for open-access artificial information options that may mimic actual information whereas anonymizing it, permitting for frictionless motion of knowledge for AI coaching and growth.
The MOSTLY AI Prize goals to place artificial information technology on the coronary heart of the AI dialog – growing consciousness of its use circumstances and supporting open information sharing as a secure, scalable resolution for innovation and analysis. Submissions for the prize shut on Thursday 3 July, with winners introduced on 9 July.
The Prize consists of two unbiased artificial information challenges: the Flat Information Problem and the Sequential Information Problem.
Flat Information is information saved in a easy desk format, the place every column represents a unique sort of knowledge. For instance, a listing of sufferers with every column holding observe of their age, gender, and peak. Sequential information entails information that happens in a particular order, capturing how issues change over time, such because the day by day worth of a inventory over the previous yr or longitudinal well being information.
The launch of The MOSTLY AI Prize follows scorching on the heels of the agency’s launch of the world’s first industry-grade open-source toolkit for the creation of privacy-safe artificial information. Entrants are welcome to make use of the MOSTLY AI toolkit to finish the problem, however this isn’t a requirement.
MOSTLY AI raised $25 million in a Collection B funding spherical in 2022, and $31 million since its launch. The agency’s international purchasers embody Citi Financial institution, the U.S. Division of Homeland Safety, Erste Group, Telefonica, and two of the 5 largest U.S. banks.
Alexandra Ebert, Chief AI and Information Democratization Officer at MOSTLY AI, mentioned: “The demand for privacy-safe information has by no means been increased. AI builders, companies, and organizations of all sizes are all clambering for entry to information as they give the impression of being to coach, tune, and deploy AI – and so they’re turning to artificial information as the answer.
“The MOSTLY AI Prize is all about placing artificial information entrance and middle. We wish individuals to grasp how it may be used and place open information sharing as a secure and accessible possibility for researchers, innovators, and builders.
We’re nonetheless on the beginning line of what AI can obtain. Open information entry is vital to unlocking its full potential – however attaining that may require the broader adoption of artificial information instruments.
“The MOSTLY AI Prize is a call-to-action for anybody with an curiosity in information and AI. We wish as many individuals as doable to get entangled, to not solely foster innovation however to extend the accessibility of artificial information technology.”
