The value of human judgment
Posted: Wed Feb 12, 2025 10:38 am
Users are also provided with an interface to track access to or movement of their data, and Macie continually monitors information usage patterns to capture indicators that data is being used or processed in an authorized manner.
This requires users to give Macie access to the contents of their S3 buckets and logs, and then after about two weeks, the service will start using that data as a benchmark for normal operations. “It classifies the content, gaining an understanding of what’s going on in the vein of, ‘I know what’s in here, I know this is sensitive information, and I know how people are using this,’ and after a couple of weeks, it starts making judgments and noticing things like, ‘This is not the normal way of using it,’” Schmidt says.
Once the system has transitioned from training to full functionality, it significantly frees users from the tedious and repetitive work of sorting out information related to accessing and using corporate data stored in the cloud.
But Schmidt says the company doesn't aim to cut norway whatsapp data out of the process, because humans are the most valuable resource any security team has. Ultimately, only humans have the ability to make informed judgments about whether suspicious-looking activity poses a credible risk to the company.
"Our goal is to develop and implement tools that allow people to think, rather than doing the same tedious work every day. The value of machine learning is that it focuses people on the things that really matter, while rejecting the things that don't matter," he said.
For example, Macie does this by alerting users to the presence of data sets that may contain personally identifiable information, providing lists of all the people who may have access to it, and noting how they may have used it in the past.
"Macie might report that one set of data accesses is anomalous and looks different from the others, and then people might ask, 'Why did this happen? Was it because the user was doing something unusual or because something unexpected happened?'" Schmidt said. "Maybe a sales person decided to download a large amount of source code for some reason, or a software developer is downloading the company's financial data into their system. Those are the kinds of situations where Macie would raise an alarm."
This requires users to give Macie access to the contents of their S3 buckets and logs, and then after about two weeks, the service will start using that data as a benchmark for normal operations. “It classifies the content, gaining an understanding of what’s going on in the vein of, ‘I know what’s in here, I know this is sensitive information, and I know how people are using this,’ and after a couple of weeks, it starts making judgments and noticing things like, ‘This is not the normal way of using it,’” Schmidt says.
Once the system has transitioned from training to full functionality, it significantly frees users from the tedious and repetitive work of sorting out information related to accessing and using corporate data stored in the cloud.
But Schmidt says the company doesn't aim to cut norway whatsapp data out of the process, because humans are the most valuable resource any security team has. Ultimately, only humans have the ability to make informed judgments about whether suspicious-looking activity poses a credible risk to the company.
"Our goal is to develop and implement tools that allow people to think, rather than doing the same tedious work every day. The value of machine learning is that it focuses people on the things that really matter, while rejecting the things that don't matter," he said.
For example, Macie does this by alerting users to the presence of data sets that may contain personally identifiable information, providing lists of all the people who may have access to it, and noting how they may have used it in the past.
"Macie might report that one set of data accesses is anomalous and looks different from the others, and then people might ask, 'Why did this happen? Was it because the user was doing something unusual or because something unexpected happened?'" Schmidt said. "Maybe a sales person decided to download a large amount of source code for some reason, or a software developer is downloading the company's financial data into their system. Those are the kinds of situations where Macie would raise an alarm."