The downside of artificial intelligence applications.

After the startup OpenAI launched its first AI application, ChatGPT, in late 2022, a wave of AI application development, especially generative AI, was triggered, bringing many benefits to all aspects of life. However, this also brings many risks.

Invasion of privacy

In recent years, many organizations and individuals have suffered significant losses as cybercriminals have used AI to create fake video clips that mimic the images and voices of real people. A prime example of this is the Deepfake scam, which involves deceiving the real person into believing they are real.

According to the Identity Fraud Report published by Sumsub in late November 2023, deepfake scams globally increased tenfold in the two years between 2022 and 2023. This was also the period when AI-generated applications exploded worldwide .

Status Labs notes that Deepfake technology has significantly impacted culture, privacy, and personal reputation. Much of the news and attention surrounding Deepfake has focused on celebrity pornography, revenge content, misinformation, fake news, blackmail, and scams. For example, in 2019, a US energy company was defrauded of $243,000 by hackers impersonating company leaders and their voices, demanding employees transfer money to partners.

Reuters reported that in 2023, approximately 500,000 Deepfake video and audio files were shared across social media worldwide. Besides Deepfakes created for fun, there are also scams designed by malicious actors to deceive the public. Some sources indicate that in 2022, Deepfake scams worldwide caused an estimated $11 million in losses.

Many technology experts have warned about the downsides of AI, including issues of intellectual property rights and authenticity, and further, intellectual property disputes between "works" created by AI. For example, someone might use AI to paint a picture on a certain subject, and another person might also use AI to do the same, resulting in paintings with many similarities.

This easily leads to ownership disputes. However, to date, the world has yet to decide on recognizing copyright for AI-generated content (recognizing copyright for individuals commissioning AI creation or for companies developing AI applications).

An image created by an AI application.

It's difficult to distinguish between genuine and fake.

So, can AI-generated content infringe on copyright? Technologically, AI-generated content is synthesized by algorithms from data they have been trained on. These databases are collected by AI application developers from various sources, primarily from the vast knowledge base on the internet. Many of these works are already copyrighted to their owners.

On December 27, 2023, The New York Times filed a lawsuit against OpenAI (with ChatGPT) and Microsoft, alleging that millions of its articles had been used to train AI chatbots and AI platforms from these two companies. The evidence presented included content generated by the chatbots at user request that was similar to or identical to the content of the articles. The newspaper could not ignore the fact that its intellectual property was being used by these companies for profit.

The New York Times is the first major American newspaper to file a copyright lawsuit related to AI. It is possible that other newspapers will also sue in the future, especially after The New York Times' successful case.

Previously, OpenAI reached licensing agreements with Associated Press in July 2023 and Axel Springer – the German publisher that owns Politico and Business Insider – in December 2023.

Actress Sarah Silverman also became involved in several lawsuits in July 2023, alleging that Meta and OpenAI used her memoir as training material for AI programs. Many writers have also expressed concern following revelations that AI systems have absorbed tens of thousands of books into their databases, leading to lawsuits from authors such as Jonathan Franzen and John Grisham.

Meanwhile, Getty Images, a photo service company, has also sued an AI company for creating images based on text prompts, allegedly through the unauthorized use of the company's copyrighted image material.

Users may encounter copyright issues when they "carelessly" use "works" they have commissioned from AI tools. Experts always advise using AI tools only for research, data collection, and suggestions for reference purposes.

On another issue, AI applications confuse users by making it difficult to distinguish between genuine and fake content. Publishers and newsrooms may be bewildered when receiving manuscripts. Teachers also face difficulties in knowing whether students' work has used AI.

The community will now have to be more vigilant because it's difficult to distinguish between real and fake content. For example, it will be hard for the average person to detect whether a photo has been manipulated or edited by AI.

Legal regulations are needed regarding the use of AI.

While awaiting the development of tools capable of detecting AI interference, regulatory bodies need to establish clear and specific legal regulations regarding the use of this technology to create original content. These regulations should clearly indicate to the public which content or works have been manipulated by AI, for example, by default adding a watermark to images that have been processed by AI.

Source: https://nld.com.vn/mat-trai-cua-ung-dung-tri-tue-nhan-tao-196240227204333618.htm