2024 Headquarters of 23andme in Seoni Bale, California. Credit-Justin Sullivan-Getty Image
teaHe has genetic data of genetic inspection company 23andme. 15 million peopleI have declared bankruptcy on Sunday night with years of financial struggle. This means that all very personal user data can be sold, and experts say vast genetic data can attract the attention of AI companies to train data sets.
Subodha Kumar, professor of Fox School of Business at Temple University, said, “Data is a new oil and this is a very high quality oil.”
But AI -related companies that want to take over 23and will take a significant risk of reputation. Many people are doing terrible things because they have given up on their genetic data to track their ancestors.
Kumar, director of the FOX ‘S Business Analytics and Disruptive Technologies Center, said, “Anyone who touches this data is taking risks.”
Read more: 23andme applied for bankruptcy. What does it mean to your account?
Education LLM
The company likes it Openai and Google In order to affect the medical field, the time and resources were spent, and 23andme’s data trobes can attract large corporations as a financial means to acquire large companies. 23andme was worth around $ 48 million This week, at $ 6 billion in 2021.
These companies are trying to build the most powerful universal models as possible, which are educated for vast amounts of subdivided data. However, researchers insisted that there is a high quality data source. dry It makes new and powerful information sauce more greedy. Techcrunch inspection More than half of the venture capitalists earlier this year, more than half of the respondents cited “quality or rare quality of monopoly data” as the advantage of AI startups in competition.
Kazlauskas said, “I think it can be a valuable data set for some large companies because it represents this evidence of the actual genetic data.”
Kumar says that 23andme’s data can be particularly valuable to the company. Agent AIAIS that can be done without human participation, regardless of medical research or company decision.
“The entire goal of the agent AI model was a modular approach. I broke the smaller problem and then arranged them,” he says.
Representatives of Google and Openai did not immediately respond to the request.
Industrial -based value
23andme data can be valued in various industries using AI.
23andme has already signed a contract. Pharmaceutical company Like Glaxosmithkline, which affects the company’s data set to develop a new treatment for the disease. Kumar said in Temple that he and his colleagues are doing projects for personalized treatment for ovarian cancer patients, and genetic data said, “It can be very powerful in understanding the structure that we can’t understand.
However, Alex Zhavoronkov, the founder and CEO of Insilico Medicine, argues that 23andme’s data will be especially worth it. “Most of the low hanging fruits have already been collected and there are important data in the open areas that are published with major academic papers.”
But many other industry companies will be interested. Anna Kazlauskas, the CEO of Open Data Labs, and Vana producer, a user -owned data network, are an abnormally large and subtle data set. “All data that suits this situation allows you to actually get valuable and difficult data.
Potentially interested industries include insurance companies that can use data to identify people with high health risks to increase their premiums. Financial institutions can track the relationship between gene markers and spending patterns in the process of evaluating loans. E -commerce companies can use data to adjust ads to those with certain medical conditions.
Ethics and personal information protection issues
But companies are also facing a significant reputation risk for participation. 23 Andme suffered hack In 2023, we exposed millions of users’ personal data to seriously harm the company’s reputation. Kumar said bidders from other industries could have much less data protection than 23andme. “My worry is that some companies are not used to having this kind of data and may not have enough governance,” he says.
This is especially dangerous because genetic information is essentially sensitive and cannot be changed once it is damaged. The genetic information of the family members of the people who are willing to provide data to the company is also in danger. Given that AI is well known prejudiceMisuse of such data can lead to discrimination in areas such as employment, insurance and loans. Friday, Attorney General California Rob Bonta Released The “urgent” warning to the 23andme customer advises the company to delete the data to the company and to destroy the genetic sample under the California Privacy Act.
EVA Galperin, a cyber security director of Electronic Frontier Foundation, is worried that 23andme’s genetic data can be in the permanent flux state of the market. “After selling data, there is no limit to the number of times the data can be reserved,” she says. This allows genetic data to fall into the hands of the organization that does not prioritize ethical considerations or can take powerful data protection measures.
Zhavoronkov of Insilico Medicine means that all these fears will be attempted by potential AI bidders to buy 23andme and its data. “Their data set is actually toxic,” he says. “Anyone who lives it will receive a negative publicity, and the acquirer will be investigated or sued.”
Kazlauskas says he is at least grateful for the mystery of having a greater conversation about data sovereignty, regardless of what’s ultimately what happens. “We’ll probably want to avoid this kind of situation that we decided to do genetic tests in the future. And five years later, the company is having a financial difficulty, and now there is a risk of genetic data to be sold to the top bidder,” she says. “In this era, the data is very valuable.”
-With the report of Billy perrigo
Please contact us ~ letters@time.com.
