Safety teams can respond 80 % faster to events using data ratios that are on behalf of Cyberhaven

Workers in institutions yearn to take advantage of the tools of artificial intelligence – whether the employer loves them or not. This useless use is known as Shadow AI, increasing significantly: as much as it increases 96 % of work Employees with artificial intelligence are through non -companies accounts. Whether it is unintentionally or harmful, this can leak from the very sensitive and ownership institutions.

Security platform Cyberhaven He says he can solve this problem by tracking data rates or data life cycle through different users and finish points. The company has Linea AI, the company, the next generation of its platform, which aims to help stop the AI ​​shadow and predict what may be accidents that may be more dangerous.

“This manifests itself in this type of lineage: you understand where the data comes from, which you managed to access, through all the different end points, through all users,” Nishant Duchi, chief product and development official at Cyberhaven, told Venturebeat in an exclusive interview.

90 % decrease in accidents that require manual review

According to the Cyberhaven analysis of the workflow of 3 million workers, the use of artificial intelligence grew 485 % Between March 2023 and March 2024. Employees are increasingly involved in sensitive data: approximately 83 % of legal documents and about 50 % of the source code, research and development materials, human resources records and employees that employees share with artificial intelligence in artificial intelligence accounts.

To help prevent unreasonable use and protect the sensitive company data, Linea Ai Libim is used on billions of actual institution data flows. Equipped with a computer and multimedia vision, it can analyze data from images, screen clips, artistic plans and other materials. The new “Let Linea Ai” feature is now independently evaluated by politics and scales of the accident to help reduce fatigue at the security operations center (SOC).

“So just like the Great Language Model (LLM) that expects the following word, we expect what will be the following procedures,” Duchi explained.

Cyberhaven claims that, as a result, customers are witnessing a 90 % reduction in accidents that require manual review and a decrease of 80 % in the average response (MTTR) for security safety accidents. The company’s tools can discover 50 decisive risks per month that are not discovered by traditional tools.

“Cyberhaaven explains to us exactly how our data moves and is used throughout the institution, allowing us to see traditional security tools,” said Prabet Karanth, CSO and COIS for the family’s financial application. Green light. “We now have a single platform not only covering the prevention of traditional data loss (DLP) and risk management from the inside, but it actually understands how people use data through our entire organization.”

Duchi explained that, while traditional methods focused on matching patterns – identifying network patterns and data to detect anomalous cases and weaknesses – Cyberhaven performs a content and context examination. The statute examines data and provides context about it based on the effects of lineage.

“If you download something, then you send it to me, and send it to five other people, they send it to five other people – this is the lineage,” Doshi explained.

How to protect Cyberhaven the most valuable data with artificial intelligence

The Cyberhaaaven view is run by FRONTIER AI models and the transformer neuron structure. “He uses a multi -stage generation (RAG) engine to analyze LLIM to analyze the most valuable institution data and” get the needle in the straw pile. ”

Aaron Arkin, chief security engineer on the library wage platform, said the platform performs a smart screen analysis, which was “a continuous blind spot.” Dailypay.

Therefore, for example, he says the security team wants to prevent screenshots from leaving the company. There may be thousands, and they must go through each one to determine whether Mimi is a harmful cat or a screenshot that contains product schemes.

“It is difficult to discover, not to mention the prevention of engineering designs, artificial intelligence models, search data, product road maps,” said Arkin.

Maintaining the tabs on users

Cyberhagen is now bearing cyber security a step beyond the disclosure of its new independent and active acting on behalf of Itea, which is decided to turn through data records and user records to help safety teams understand the severity of the accident. Dochi explained that the statute understands screenshots, PDFS, source code and other digital materials and can provide a context based on data rates. It is then possible to distinguish whether a specific incident is needed by human analysts.

“We are trying to predict the following procedure based on all the historical knowledge that we have: this is an anomaly, or this is a benign event,” said Duchi. “We call this data understanding, because you are really looking for data and understanding that in -depth data.”

Arkin explained that when it comes to internal dangers, safety teams are monitoring improved to create information flows about specific users who have been marked as an increased risk (based on any number of factors).

“Let’s say I put a reinforcement on you, I was busy today, 150 events were created,” he said. “I must go to each of these manually, select“ this work as usual. ”This looks a little suspicious.” This looks really suspicious. ”I still have others after that.

For example, the platform has been able to discover users who send data to OneDrive accounts or synchronize sensitive files with iCloud, said Doshi. The malicious step that exceeds this is to leave employees a company and try to take sensitive data with them.

“We can in an actual time to prevent users or a group of users from downloading sensitive data to this general LLMS,” Douche said. “We can also warn them and educate them” when they do something unintentionally or naively.

Arkin said, for its part, Dailypay managed to reduce MTTR by 65 % because Linea provides an AI summary. DLP loss tools require a lot of employee resources to obtain this type of vision.

I have searched for the other DLP providers including Netskope, DTEX Systems and the Next DLP, but eventually settled on Cyberhaven due to the data rate strategy. He said he was unlike anything he saw in the industry.

“It provides us with a lot of time on escalation, trilogy and prevention as well,” said Arkin. “Linea AI constantly determines the accurate risks that traditional systems will completely miss them.”