Wondering whether your data will be used for AI training? Researchers have developed so-called copyright traps to find out exactly that.
It’s like in sports: behind a good AI model there is also good training. But artificial intelligence requires huge amounts of data for this training. However, many authors view this critically because they may not want companies to use their content or works to train AI models without consent.
Researchers at Imperial College London have now developed a way to unmask exactly this data from AI training. These are so-called copyright traps, which, so to speak, set a trap for the AI.
What data does AI use for training?
Copyright traps are nothing new for copyright compliance. But now they can also be used in the field of artificial intelligence.
Yves-Alexandre de Montjoye, a professor at Imperial College London who led the work, presented the results at the International Conference on Machine Learning. “There is a complete lack of transparency around what content is used to train models, and we believe this prevents there being a true balance between AI companies and content creators,” explains the scientist.
How do copyright traps work?
The way these traps work is quite simple. For example, authors can hide a piece of text in a data set that actually makes no sense at all. If an AI model uses this later, it becomes apparent that the data set was used for AI training.
The team at Imperial College London has developed sentences that look like this in English: “It’s my favorite time of the year: the time between New Year’s and Easter; there are so many.” Translated, this means something like: “It’s my favorite time of the year: the time between New Year and Easter; there are so many”.
This is how you can use copyright traps
If you also want to use such a trap, you can at GitHub find what you are looking for. Copyright traps for Large Language Models are already available there. These provide you with the script and also generate text traps for checking AI.
However, this is likely to become even easier in the future. Because the team around Yves-Alexandre de Montjoye is working on a tool. Authors should then be able to use this to create copyright traps in order to integrate them into their texts.
Also interesting:
- GDPR violation? X collects user data for AI training – without consent
- Reddit blocks search engines: Is the time of the open internet over?
- Study reveals: This is how you can distinguish deepfakes from real recordings
- AI could replace 165,000 civil servants in Germany
The post Copyright Traps: How to find out whether your data is being used for AI training by Maria Gramsch first appeared on BASIC thinking. Follow us too Facebook, Twitter and Instagram.
As a Tech Industry expert, it is crucial to be vigilant about copyright traps when it comes to AI training data. One of the key ways to find out whether your data is being used for AI training is to thoroughly review the terms and conditions of any agreements or contracts related to the data you provide.
It is important to ensure that you are the rightful owner of the data being used and that you have the proper permissions to use it for AI training. Additionally, conducting regular audits and checks on the usage of your data can help identify any unauthorized or questionable activity.
Using watermarks or other tracking measures on your data can also help monitor its usage and prevent unauthorized use. Collaborating with reputable and trustworthy partners for AI training can also help mitigate the risk of copyright traps.
Overall, staying informed and proactive about protecting your data rights is essential in the constantly evolving landscape of AI technology. By being aware of potential copyright traps and taking proactive measures to protect your data, you can ensure that your information is being used ethically and legally in AI training processes.
Credits