LinkedIn, the professional social media platform owned by Microsoft, has developed an AI that can answer questions about whether products, data downloads, and other digital tools comply with a company’s security policies or pose cybersecurity threats.
The AI, now in testing, can return answers within five seconds, compared to 15 minutes when humans are tasked with similar questions, and is 90 percent accurate, the company says.
In addition to saving large amounts of time and effort across a business, the AI can heighten security against hackers, malware, and other ills, according to Jeff Belknap, LinkedIn’s security chief.
When the clearance process takes more time, impatient employees can skip steps in vetting an item’s security, he said. By automating the process and condensing it to a few seconds, no steps are missed, he pointed out in a presentation on the new AI.
Similarly, the new security AI speeds the process of collaboration between companies. Normally, security specialists within companies have to exchange lengthy questionnaires about their security processes with potential corporate partners. The new AI can automate the mutual examination to determine compatibility before the day is over.
When humans carry out the process, “an inconsistent or incomplete answer might result in a deal being delayed for weeks or months or falling through,” Belknap said.
The company is still testing the tool to make sure it’s not falling victim to “AI disease” of giving false information and blurting racially or otherwise offensive remarks.