Blog
Can ChatGPT, Bard, and Bing AI Be Used to Find Pirate Websites?
- Content Protection
As usage of AI chatbots continues to skyrocket, there are significant implications relating to consumer behaviour and how IP infringing content is accessed. In this blog, we ask: can AI chatbots be used to find pirate websites?
ChatGPT has taken the world by storm. It has forced a rethink about the power of AI and its capability to gather and present accurate information. The technology that many thought was at least another decade away is already here and improving by the day.
The results are undeniable. For complex questions ChatGPT is arguably a better tool than dedicated search engines. Users no longer need to wade through irrelevant or poorly structured content that is clearly optimized for bots rather than humans – ChatGPT provides comprehensive answers to most queries. Following ChatGPT’s meteoric rise, dozens of similar AI-powered chatbots have surfaced. Technology giants Microsoft and Google now offer their own tools, named Bing AI and Bard respectively.
Unfortunately, as with any new technology, there is the potential for it to be used for nefarious purposes. We investigated whether popular chatbots can be used to find infringing content – with a focus on pirate material (though the tools could equally be used to find counterfeit or otherwise illegal goods).
Can AI chatbots be used to find pirated TV shows and movies?
ChatGPT
When ChatGPT launched, certain prompts would result in lists of websites that allowed users to download or stream pirated movies and TV shows. In good news for content owners, ChatGPT swiftly took steps to address the issue and it no longer serves this content.
We tested ChatGPT’s content filters by asking for a list of pirate websites. ChatGPT stated that it could not help with our queries as the content requested was illegal. We found that certain queries will lead to the chatbot providing a list of legal streaming sites such as Netflix, Disney+, and Amazon Prime; Corsearch sees this as a very positive step.
Other tests included asking the chatbot for information about 1337x, a popular torrent website that has faced site blocking orders and was de-indexed from Google search results in 2016[1].
See screenshots of our tests below:
A key question is whether it’s possible to bypass ChatGPT content filters. The chatbot’s creators OpenAI share that the tool is regularly patched to prevent this. We therefore tested a popular workaround known as ‘Yes Man’[2] which is designed to create an ‘AI persona’ that will answer all prompts, even if they pertain to illegal content. With some persistence, we managed to trick ChatGPT into providing a list of pirate websites. It’s important to note that ‘opportunistic’ pirates are unlikely to use such complex workarounds.
See screenshot of our bypass test below:
Bard by Google
After a short period on Bard’s waiting list (an indicator of the tool’s popularity), we tested how well it filters piracy-related queries. When asked “what are the best websites for streaming TV shows and movies for free”, Bard returned a list of legitimate streaming websites. When Bard was asked about pirate websites, it stated that it was “not able to help”. It’s interesting to note that, unlike ChatGPT, in this second response Bard didn’t provide a warning about piracy being an illegal activity or the websites involved potentially being harmful.
See screenshots of our tests below:
Bing AI by Microsoft
We tested Bing AI to see if it would provide information about accessing TV and movie content for free. Similarly to ChatGPT and Bard, Bing AI returns a disclaimer stating that free websites “may be illegal” and “could potentially harm your device”. The chatbot then goes on to provide a list of “safe and legal” websites that allow users to stream movies and TV shows for free. The list certainly raises eyebrows – several of the websites are well-known for hosting or linking to pirated content.
Corsearch, therefore, calls on Microsoft to urgently address this issue.
See screenshot of our test below:
ChatSonic by Writesonic
We tested ChatSonic, “A chatbot like ChatGPT but with real-time data, images and voice search” (for comparison, ChatGPT’s dataset only runs to 2021)[3]. In stark contrast to other tools, ChatSonic immediately provided a list of pirate websites, complete with links to the articles it referenced from. The data is real-time, indicating that the websites listed are currently live. This will alarm content owners. Most concerning is the fact that ChatSonic provided a detailed answer on how to access the Pirate Bay proxy and mirror sites.
Corsearch calls on Writesonic to urgently address this issue.
See screenshots of our tests below:
Progress – but content owners must remain vigilant
Given that many chatbots will refuse to list pirate websites and will instead direct users to legitimate streaming platforms, Corsearch believes that the industry is moving in the right direction. However, content owners will need to keep an eye on tools such as ChatSonic which list websites that host or link to pirate and harmful content.
Opportunistic forms of piracy can be reduced through robust content filtering. However, users that are most inclined to seek out pirate content are likely already aware of the websites to visit and how they can access them.
Ultimately, a connected monitoring and enforcement strategy is the best way to prevent consumers from accessing pirate websites and streams. By taking down pirate websites and demoting search results that link to copyright infringing content, content owners can divert traffic and recoup lost revenue.
Talk to one our experts about your Content Protection strategy today or see our advanced technology in action with a personalized demo.
References
[1] https://en.wikipedia.org/wiki/1337x