[ITmedia Business Online] Why doesn't Zoho "advertise"? Ask the AI research director about his “unique design” to prevent information leaks

Confidential information has been leaked one after another at companies. For Japanese companies, where the introduction of generative AI is accelerating, information leaks are no longer just a fire on the other side.

According to the Information and Communications White Paper by the Ministry of Internal Affairs and Communications, among the risks that companies are concerned about when using generative AI, the highest proportion of respondents answered ``security risks such as leakage of internal information.'' Meanwhile, a survey by Kadokawa ASCII Research Institute found that 46.5% of managers have used AI tools in their work without the company's permission. Many companies are caught between improving operational efficiency and protecting data.

Zoho, an Indian company that produces business software such as CRM (customer management) and provides services to over 1 million companies and organizations around the world, does not even have advertisements on its free version. Not only does it not sell user data to outside parties, but it also maintains a design in which "customer data is never used for AI learning."

How can you protect your company's data, which is its lifeline, without stopping the use of AI? We asked Ramprakash Ramamurthy, head of the company's AI research department, about this.

What is the purpose of not using customer data for AI learning?

--Zoho takes the position of ``absolutely protecting data privacy.'' Giant US IT companies, including GAFA, use user data to train AI. Recently, there have been incidents in Japanese companies where information was leaked to outside parties. What is the basis for saying that Zoho can protect data?

We have been very particular about data privacy since long before we adopted AI. Our products do not display any advertising. The same goes for the free version. We do not have any trackers (mechanisms that track visitor behavior) installed on our own website. If you look at zoho.com, you'll see that there are zero trackers. One of our strengths is a design that does not create any chance for data to be leaked to the outside.

Information for each company is that of that company. For example, an insurance company might use its own data to use AI to predict the best time to contact customers. We do not use this data at all for our own research or AI training. If a competitor of that insurance company starts using our product in the future, there is a risk that the AI ​​design that the insurance company has built using data will be leaked to the competition.

For this reason, we have strict provisions in our contracts. It does not handle data in the customer's environment, nor does it have anything that would create an opportunity to do so. We are very determined not to retrieve any data. Our starting point is the idea of ​​protecting data privacy, with or without AI, and that's what drives our robust designs and products.

--OpenAI, which provides GAFA and ChatGPT, uses large amounts of data for AI learning. What do you think of these movements?

I think the premise of that comparison is a little off. OpenAI also has accounts for businesses, which are handled differently from those for individuals. In the first place, it is common sense not to handle company information in personal accounts.

Data from personal services should generally be used for learning everywhere. Even when you simply look up a location using a smartphone's map function, that information is circulated and used as data that helps companies develop profitable products.

I don't think they can be compared on the same level as accounts for businesses and services for individuals are treated differently.