AVIS Car Rental Breach Exposes Sensitive Customer Data
September 9, 2024
Menu
This is fairly obvious given that LinkedIn has to safeguard a potentially enormous number of users since their customers worldwide account for over one billion individuals. In the face of ever more cybersecurity attacks, significant challenges to manage vulnerabilities were apparent. LinkedIn developed its own AI solution to help protect their hardware and users.
For LinkedIn, the process started by establishing a canonical hardware system of record That is, bringing data from many sources into one place and confirming that the details are both correct and current. To surface vulnerabilities and forecast potential attack paths, LinkedIn constructed the Security Knowledge Graph so it could maintain real-time information on its assets.
The social network LinkedIn has teamed up with Microsoft, utilizing sophisticated AI to get the security team ready for what may come in terms of vulnerabilities. Its aim was to surface and eliminate security issues for every type of discipline — especially engineering, and LinkedIn, rather than going with a traditional API decided to use the GraphQL API. This enabled the engineers to zero in on relevant information by following trails within various data nodes.
It can also learn from previous queries, improving its answers over time as it becomes more familiar with common issues and concerns. If a question does not generate the right answer, other measures are taken to ensure that the team can source their required information.
Credential theft is one of the most prevalent threats in cybersecurity. LinkedIn went a step further, specifically restricting access to its machine-learning system. Only a trusted group of engineers can access it, and all queries are monitored for unusual usage patterns. The system will most likely be able to determine if a hacker is trying to enter by how they question.
How LinkedIn does AI to support its team When building AI products at LinkedIn, we also think about this idea that Sabry Tozin VP and Head of Engineering for Careers moved forward: “AI is not here to replace engineers — it has been designed to make engineer lives more productive. Artificial Intelligence assists in faster and more accurate insights, however humans are required to verify all the information.
The AI project at LinkedIn started several years ago using an earlier version of GPT technology. In the beginning, liquidation was only 40%–50%, but in this latest GPT-4 generation, they have increased it to up-to-date accuracy of around 85%-90%. This improvement was a result of ongoing tweaks by the AI to reduce mistakes and make better suggestions.
State-of-the-art cybersecurity has seen an addition in the form of LinkedIn’s AI-powered vulnerability management system. With its stable and trustworthy system in place, LinkedIn can more rapidly find its vulnerabilities before they are attacked and provide engineers with fast accurate answers on demand. New technologies like this one keep LinkedIn a step ahead of increasing cybersecurity threats and secure all its users.
LinkedIn uses this real-time storehouse to keep track of its hundreds of thousands digital assets, and the security team references it constantly to find vulnerabilities before a breach can happen.
LinkedIn uses AI to drive a higher velocity SecOps approach AI navigating knowledge graph Q&A for the security sector.
Yes, LinkedIn’s AI system is tightly controlled. Only a small group of security engineers can access it, and all activities are monitored to detect any unauthorized attempts.
AI at LinkedIn is used to augment engineer productivity and not replace them. All information is human-verfied, AI just helps analyze the data.
LinkedIn has been testing and refining its AI over time, with early tests yielding an accuracy of 40%-50%, increasing to approximately 85%-90% at the GPT-4 level as it stands now.
