AI-Generated content detection
Telefónica Digital Innovation is working on different solutions to detect AI-generated or manipulated content to help society navigate the new world of artificial media.
We stand at the cusp of a new era that will profoundly change our media experience, blurring the boundaries of reality and deception. Generative AI tools like ChatGPT, Midjourney, Dall-E, Runaway.ai, or Jasper are now in the hands of millions, enabling the creation of synthetic content that is nearly indistinguishable from human-made material, potentially causing harm to individuals and society at large.
People taking a photograph
Helping citizens navigate the new world of artificially generated media
Motivated by the urgency of the situation, various teams at Telefónica Digital Innovation are researching, exploring, and developing different means to detect a wide range of AI-generated content, as its impact can be far-reaching:
AI-generated images and videos have reached a point where distinguishing them from real photographs is increasingly challenging. Deep fakes in videos can put false statements into people’s mouths or make them appear in situations they have never been in. These manipulations can lead to devastating consequences such as the spread of fake news, conspiracy theories, discrediting innocent individuals, election manipulation, and incitement of outrage, unrest, and violence. Our democratic societies face serious threats when we can no longer trust our own eyes.
AI-generated text: AI-generated text is another area of concern, with Large Language Models like ChatGPT, Jasper, or Bard capable of producing high-quality content that can be exploited for cheating in educational settings, posing a significant threat to our education and academic integrity, as we know it today.
Real-time AI voice generators are becoming increasingly sophisticated, enabling impersonation of individuals with only a few seconds of their voice. Detecting AI-generated voice can be pivotal in preventing fraud in phone calls and voice messages.
AI beauty filters may contribute to unrealistic beauty standards, impacting particularly young people's mental and physical health as they struggle to match the edited images they see on Social Media and share with the world themselves.
The most significant challenge we face is the rapid evolution of generative AI tools. New models, tools, and services emerge every week, necessitating the development of detection methods that can keep pace with this constant evolution.
Our mission is crystal clear: to help citizens navigate the new world of artificially generated media, promoting clarity, safeguarding mental health, and strengthening our democracies.
