TechDogs-"What Are The Concerns Around Visual Replication In Generative AI?"

Emerging Technology

What Are The Concerns Around Visual Replication In Generative AI?

By TD NewsDesk

TD NewsDesk

Updated on Wed, Jan 10, 2024

Overall Rating
You've probably seen those vivid, imaginative images popping up on social media created by AI systems such as DALL-E and Midjourney. They can whip up original visuals from simple text prompts. Yet, what if these Art Generators are actually just repurposing other artists' work without permission?

TechDogs-"Screenshots Of The Study Which Shows All The Images Generated By Midjourney That Appear To Be Copyrighted Material."
The extent to which large language models (LLMs) absorb and "memorize" the training data has sparked ongoing debate among scholars, with luminaries such as Google DeepMind’s Nicholas Carlini and Gary Marcus, involved in the matter. Recent studies have shed light on LLM’s visual plagiarism problem!

What Did They Say?

  • In a noteworthy example, a 2023 research paper authored by Milad Nasr and colleagues demonstrated how LLMs could be persuaded into divulging sensitive information like email addresses and phone numbers. 
  • Additionally, DeepMind’s Nicholas Carlini and his collaborators revealed instances where larger chatbot models reproduced extensive text verbatim, a phenomenon less common in smaller models.
  • The recent legal clash between The New York Times and OpenAI further showcased instances where the latter's software mirrored New York Times articles almost word for word, raising concerns akin to what would conventionally be labeled as plagiarism if produced by a human.
  • Such revelations raise concerns that certain generative AI (Generative Artificial Intelligence) systems may generate plagiaristic outputs inadvertently, potentially exposing users to copyright infringement risks.
  • It turns out Midjourney has a "visual plagiarism problem" that can too easily recycle elements from existing images when generating new ones. Visual artist Daniel Southen recently put Midjourney V6 through its paces and uncovered this dark secret.

So, Who Is Ultimately Responsible If AI-generated Art Violates Copyrights?

  • The stakes around Artificial Intelligence (AI) and copyright only seem to grow trickier. As language models like GPT-3 and GPT 3.5 have already demonstrated, neural networks tend to absorb anything they're trained on, including passages from novels and news articles. 
  • As long as the legal lines remain blurry, lawsuits from targeted artists could accuse AI businesses, like Anthropic who developed Midjourney of plagiarizing their images. Whether an AI system could be sued remains an open and intriguing question that may ultimately need to be settled in court.
  • Surprisingly, 83% of organizations agreed AI is a top priority in their business plans!

Perhaps such tech companies should implement stronger filters on training data in a bid to curb copying. However, crafty AI seems to always find new ways of connecting dots that humans can’t anticipate.

While AI's imitative capacities underpin plagiarism worries, experts also highlight how humans intentionally built neural networks to find new links in data. Some degree of "automated creativity" may necessarily involve mining copyrighted works used in training. In other words, a form of “accidental” AI plagiarism can be taken as a pinch of salt in the AI generated content.

Striking the right balance between innovation and intellectual property protection stands out as a primary challenge for advanced AI systems. Updated regulation and development practices can tackle AI-enabled visual plagiarism and further uphold creative rights in the digital age.

Do you think it is possible to get rid of the generative AI’s visual plagiarism problem? What steps can AI companies take to ensure their systems reduce visual replication?

Feel free to drop your thoughts in the comments section below!

First published on Wed, Jan 10, 2024

Enjoyed what you've read so far? Great news - there's more to explore!

Stay up to date with the latest news, a vast collection of tech articles including introductory guides, product reviews, trends and more, thought-provoking interviews, hottest AI blogs and entertaining tech memes.

Plus, get access to branded insights such as informative white papers, intriguing case studies, in-depth reports, enlightening videos and exciting events and webinars from industry-leading global brands.

Dive into TechDogs' treasure trove today and Know Your World of technology!

Disclaimer - Reference to any specific product, software or entity does not constitute an endorsement or recommendation by TechDogs nor should any data or content published be relied upon. The views expressed by TechDogs’ members and guests are their own and their appearance on our site does not imply an endorsement of them or any entity they represent. Views and opinions expressed by TechDogs’ Authors are those of the Authors and do not necessarily reflect the view of TechDogs or any of its officials. All information / content found on TechDogs’ site may not necessarily be reviewed by individuals with the expertise to validate its completeness, accuracy and reliability.


Artificial Intelligence AI Generative AI Large Learning Models DALL-E Midjourney OpenAI ChatGPT

Join The Discussion

  • Dark
  • Light