New AI Models Detecting Threats of Violence, New Image Inspection for Chat & Email Attachments, Simplified URL Reporting and More
New Assure Features - July 2023
Introducing the latest version of Assure, and our biggest release so far this year. New AI models targeting threats of violence, extremism and radicalisation, simplified URL reporting, and image analysis across student chat and email.
Read on to see the highlights of Assure v.5.16
New AI Models Targeting Threats of Violence, Extremism and Radicalisation
Saasyan Assure's Safe Image AI and natural language processing engine have helped hundreds of schools detect and intervene against cyberbullying, self-harm, and sexually explicit content across students chats, emails, and online drive accounts.
Today, Saasyan is taking this even further and expanding Assure's Natural Language Processing Engine to include threats of violence, extremism and radicalisation as well as our brand new, Image Object Detection AI.
Natural Language Processing Model for Threats of Violence, Extremism and Radicalisation
Assure's natural language processing engine (NLP) is a vital part of how we deliver language-based alerts. The NLP is applied to students' chats, emails, drive content, searches and even video titles to provide alerts beyond direct matches to phrases found in Assures local 'words & phrases' dictionary.
For example, if a student sends a message that says, 'you are so annoying, go home', none of these words would probably be found in Assure's local dictionary, however given the context, our NLP would flag this as cyberbullying.
In Assure Version 5.16, the NLP engine now includes a new model for threats of violence. This means if a student performs a search for something like 'Where to buy a gun' or ‘best place to hide a knife’, you will receive an alert regardless of if the trigger terms are found within the local dictionary.
Image Object Detection AI
With the launch of our Safe Image AI last year, Assure redefined the way schools target image-based abuse with the ability to identify images containing sexually explicit content within Microsoft 365 and Google Workspace.
Assure version 5.16 is expanding this further to directly target threats of violence, extremism, and radicalisation with our Image Object Detection AI.
Capable of identifying more than 4,000 different types of weapons, if an image anywhere within Microsoft 365 or Google Workspace contains an object that is identified as a weapon, your school will know.
Simplified URL Reporting
Assure's comprehensive online activity reports are fundamental in enabling schools to investigate and follow up on concerning incidents, and overall student online activity.
One of the key features in online activity reporting, is URL based reports, which display a user or group's website activity.
In Version 5.16, Assure now intelligently groups website activity to provide a single entry displaying the base URL for each website hit. For example, if a student watches several videos on YouTube and the base URL is different for each page, these will now be compiled within a single entry.
If a user wants to examine the full website activity beyond the base URL, they can simply click on the base entry which will display an expanded view with the entire list of sub URLs for that website hit.
This makes Assure's reporting functions even more accessible for wellbeing staff, school leadership, and classroom teachers to access a user's web activity and easily differentiate the important signals from unnecessary noise.
Image Inspection and Analysis for Chat and Email Attachments
Each week, thousands of images are shared across students’ chat and email accounts. Today, we are expanding Assure’s capability to inspect images within One Drive and Google Drive, and bringing this directly to student chat and email.
By applying Saasyan’s Safe Image AI and our newly announced Image Object Detection AI, Assure will now be able to identify image-based abuse, nudity, sexually explicit content, and more than 4,000 different weapons across students’ chats, emails, and online drive accounts.
When an image-based alert is triggered from a chat or email thread, users will have two options to investigate. One is to investigate the image that has triggered the alert directly and the other is to investigate the image within the context of the conversation.
When displaying the context of email alerts, Assure intelligently compiles all of the images found within an email thread in a consolidated view.
For alerts generated within Microsoft Teams Chat, the image itself will also be accessible as an attachment within the message for easy investigation.
Get in touch with our Support Team.