Researchers From MIT May Have Found the Holy Grail of Network Security

When it comes to cybersecurity, maintenance is key. Whether you choose human-based security or an automated security solution, running into shortcomings is still possible. Human security tends to rely on the word of experts, and anything that doesn’t fit into the guidelines is missed and may therefore get through and wreak havoc. Network security can be a touch overzealous, in a way “crying wolf,” with an excess of false positives that ultimately require human analysis, leading to human frustration.

Blending the two in the past has proven difficult, as the experts with the necessary skills aren’t usually available for the time it takes to label the data for the programs to work properly. However, a team of researchers assembled from MIT’s CSAIL (Computer Science and Artificial Intelligence Laboratory) a PatternEx (a machine-learning startup) recognized these issues and have developed a platform that blends AI and human security attributes in a way that shields from the shortcomings of either.

Dubbed A.I.², this new platform is capable of detecting 85 percent of incoming attacks while reducing reported false positives to 20 percent of what they were originally. To achieve this, the “untrained” machine reports what machine learning tells it are the 200 most important problems in the sample set to a human expert, who then corrects the machine’s work. As the machine progresses through sets of data, it reports fewer and fewer false positives.

In doing so, A.I.² demonstrates machines effectively learning from human teachers, cooperating for the sake of security analysis accuracy.

The implications to the realm of security in the future are massive. It’s difficult not to let the imagination run wild with thoughts of upcoming technologies being influenced by this human-artificial intelligence hybrid.

However, since there is no telling how long it may be before a system like A.I.² could be available to users for purposes of data defense, business owners should still follow best practices. Users should not only be educated on the importance of avoiding typical security risks. They must also be reminded of the importance of compliance to regulatory standards for business security, such as those set by organizations like HIPAA, DSS, and others. They should also be encouraged to review the reports generated by security software.

Additionally, you could always allow Graemouse Technologies to help manage your cybersecurity. By intelligently implementing security solutions, we can help your company avoid security issues and keep your IT in top working order. To set up a remote monitoring and maintenance solution, call 253.777.0763 today.