Hundreds of new items of extremist and illegal content are being created online daily
The amount of extremist content, terrorist propaganda and hate speech being posted online, especially on social media platforms is a growing global issue. World leaders have challenged the world’s biggest technology companies to remove extremist and illegal content in a timeline as short as under two hours.
Crisp is now deploying its advanced Artificial Intelligence with major social networks across the globe to identify terrorist and extremist content online. Our technology – called Capture – has been locating nearly 200 pieces of new, unique pieces of terrorism-related content every day. This is content that has been known to have a ‘starburst’ effect, where it is rapidly shared, distributed and viewed many thousands of times.
Crisp's Capture technology has identified 5 key categories of illegal content. Of the content identified:
73% non-distressing terrorist propaganda
Images and videos which glorify or promote the terrorist cause, but without the use of extremely distressing material
15% violent terrorist propaganda
Beheadings and bloody gore resulting from terrorist activity shown in the context of glorifying or promoting the terrorist cause
8% contains extremist preaching
Videos and written pieces that promote hatred or violence against others
2% instructional videos on terrorist devices
Detailed instructions on how to create explosive, chemical devices or weaponry for use in terrorist acts
2% incitements to commit terrorist acts
Content that directly incites the audience to carry out terrorist acts that would result in the harm or death of members of the public
Capture is a revolutionary new service
Capture works very differently to traditional ‘seek and destroy’ moderation methods often used by social networks to date. Due to the vast amount of content uploaded every minute to social networks, trying to find extremist and illegal content online amidst all legal content is akin to finding a needle in a haystack.
Capture works by graphing the social web, finding the places that have been identified by the AI as risk hubs for the discussion and sharing of illegal content, such as in dark web forums, discussion boards and messaging apps, and following those discussions to locate the source, so it can be reported and removed, within minutes of being discovered.
Capture can identify the highest risk and most shareable content within minutes, reporting it directly to the social networks teams to action or even automate complete take down.
27% of content includes violent, distressing or incitements to commit terrorist attacks
Spikes in activity seen around key acts of terrorism
Advanced artificial intelligence technology being used to detect illegal content for leading social networks
“Our approach works on the premise that illegal content is being uploaded because the author wants people to see it. So instead of searching through huge quantities of good content in the hope of finding the bad, we monitor at scale the vast data sources that promote bad and illegal content, giving us incredible accuracy, real-time detection and scalability.”
Adam Hildreth, CEO, Crisp
Crisp’s other services
Crisp’s mission is to deliver complete safety for your social platform by providing expert services.
Image Classification and Moderation
Crisp moderates millions of images and videos in real-time every month. We offer the industry’s highest detection rate for offensive and illegal imagery, for example pornography, gore, guns, abuse and extremist content. We deliver custom risk categories such as ad compliance, dating profiles and image relevance, and manage illegal image reporting to authorities. Our global team proactively classifies millions of images and videos for training, keeping us ahead of the game.
Comment Classification and Moderation
Crisp analyzes posts, comments and discussions in over 50 languages to identify a broad range of risks from offensive, illegal and abusive content, to spam and online scams. We offer pre- and post- moderation services, and use a combination of natural language processing and machine learning techniques together with real-time, on-demand human curation for marginal decisions.
Crisp protects live chat applications with real-time contextual filtering. Our service combines advanced linguistic analysis with real-time reputational profiling. The type of content allowed and the way it is used depends on the criteria you set. We have pre-configured policies for a wide range of audiences – from ‘edgy’ to the most stringent COPPA compliant kids’ sites. Crisp’s live chat filtering services are always used alongside ‘bad actor profiling’ allowing us to determine the reputation and behavior of users, and identify risks such as sexual exploitation, trolling, radicalization and grooming.
Bad Actor Profiling and Reputation Management
Crisp has numerous patents granted in the real-time profiling of ‘bad actor’ behavior, for example abusive individuals, trolls, sexual predators, groomers and extremists. Our machine learning platform is constantly trained by Crisp’s experts with real world data from law enforcement, security and psychology experts, so we stay one step ahead in detecting fraudulent and toxic behavior. Our social media risk experts are ready 24/7 to review, manage and escalate serious issues, and impose user sanctions within minutes.