PornHub tests AI to deter users from illegal content

The British porn site PornHub has begun testing an automated tool that ‘discourages’ users from searching for materials involving child abuse (CSAM). This is reported by Wired.

Since March 2022, whenever a query containing words or phrases related to illegal content is entered, the platform launches a dedicated chatbot. It interrupts the attempt to search for CSAM materials and offers users help regarding their behaviour online. In the first 30 days of testing, the tool was activated 173,904 times.

Susie Hargreaves, chief executive of the Internet Watch Foundation (IWF), said efforts should be made to prevent the problem from arising.

“People must not search for materials about sexual violence against children, but stop and check their behaviour,” she said.

The chatbot was developed by the Lucy Faithfull Foundation and the IWF. They trained the system to search for any of 28,000 terms related to illegal content, and millions of their combinations.

On activation, the tool will ask users a series of questions and warn that the requested information may be illegal. It will also offer ‘confidential and non-judgmental’ support. Clicking the ‘help’ prompt will provide detailed information about the Lucy Faithfull Foundation, the organisation’s phone number and its email address.

“The goal is to distract a person or derail their desire to search for CSAM materials, and do so in just a few clicks,” explained Dan Sexton, the IWF’s chief technology officer.

According to Donald Findlater, head of Stop It Now, the use of the chatbot is straightforward and perhaps more engaging. After the tool’s March testing, 158 people visited the helpline. He noted that while this figure is modest, it marks an important step.

Representatives for the organisations said PornHub ‘volunteered to participate’ in the project and is not being paid for testing the system. It is expected that the trials of the chatbot on the British site will continue for another year, before being assessed by other researchers.

“The IWF tool is another layer of protection that informs users that there is no illegal content on the platform, and directs them to Stop It Now for behavioural change,” a PornHub spokesperson said.

As reported in June, Australian authorities asked citizens to share their childhood photographs to teach AI to detect abuse in photographs.

In July, the UK government backed the idea of scanning users’ smartphones for CSAM materials.

In August 2021, Apple deployed AI to check photos on the iPhone for signs of child abuse.

Subscribe to ForkLog news on Telegram: ForkLog AI — all the news from the world of AI!