New research warns of risks from data policies of AI companies

Cát Tiên |

A new study suggests that many AI companies are using misleading designs to make it difficult for users to refuse to share personal data.

A new study by the US Electronic Privacy Information Center (EPIC) suggests that many AI companies are applying manipulative design tactics to make it difficult for users to refuse to share personal data.

According to a report released by this unit, major language model providers (LLMs) such as OpenAI, Google and Meta are said to use many methods similar to traditional data brokers to reduce user data control capabilities.

EPIC said that many companies do not clearly set forms for refusing to share data on their homepages or in their security policies.

Some platforms also require users to fill in multiple different forms for the same request, making the process complicated and time-consuming.

The study also suggests that OpenAI currently does not provide a clear tool for users to refuse to sell or transfer personal data.

Instead, this platform mainly allows users to request the deletion of information appearing in ChatGPT's response, instead of deleting the platform data that the company is storing.

According to EPIC, this issue is particularly worrying because AI companies are currently participating in a vast data ecosystem where personal information can be shared between many different parties such as data brokers, advertising platforms and technology businesses.

This organization warns that the lack of mechanisms to deny transparency can lead to many serious safety risks.

The report cites a number of past cases where commercial data was used to locate, harass or target women and other vulnerable groups.

In the study, EPIC checked the privacy protection processes of 38 large data companies. The results showed that many forms of "manipulated design" are commonly used.

Some companies are said to hide links that refuse to share data in small, hard-to-find text. Others require users to create an account or pay a registration fee before exercising their privacy rights.

Researchers also found that on platforms such as X, OpenAI or Tinder, users must log in to their accounts first to be able to send requests to refuse to share data.

In addition, TikTok and Amazon are said to provide privacy forms but do not have clear options to refuse the sale of personal data.

For dating apps, research shows that Bumble users are defaulted to participate in data sharing through pre-tagged options, which can cause confusion for users.

Responding to the study, Amazon said that the company does not sell customer personal information to third parties, so users by default do not need to perform data sharing refusal.

Meanwhile, OpenAI affirms that they only share limited data with marketing partners for advertising and provide many ways for users to control privacy.

However, EPIC believes that even if the forms are improved, users will still find it difficult to protect their privacy if they have to send individual requests to each company holding their data.

According to this organization, a more effective solution is to limit data collection from the beginning and build stricter regulations to prevent businesses from storing unnecessary personal information.

Cát Tiên
RELATED NEWS

Using AI to create fake data, researcher may be banned from operations for a limited time

|

Plagiarism, data falsification or the use of artificial intelligence (AI) can cause individuals to be banned from participating in research activities indefinitely.

AI chatbot makes researchers face a crisis of trust

|

AI chatbots are causing researchers to face a crisis of confidence as nearly 150,000 fake quotes are discovered in many global scientific articles.

Claude expands drug and material research capabilities with AI

|

Claude's ability to research drugs and materials with AI is expanded through cooperation between Anthropic and scientific AI company SandboxAQ.

The Secretariat disciplined and expelled Mr. Nguyen Duc Thanh and Diep Van Chien from the Party

|

The Secretariat decided to impose the disciplinary action of Expulsion from the Party on Mr. Nguyen Duc Thanh, Diep Van Chien.

Gold ring prices fall, people take advantage of buying wedding gold to save money

|

On May 26, gold prices fell, many people in Hanoi rushed to buy in, trading faster, no longer crowded like before.

Power outage in hot weather, many households in the middle of the night find places to sleep

|

Hanoi - On the evening of May 25, some locations in the capital Hanoi experienced power outages for many hours, especially in suburban communes.

Illegal pig gathering point causes pollution in residential areas in Lam Dong

|

Lam Dong - Wastewater from a pig collection point in Quang Son commune is discharged directly into the stream, causing environmental pollution and affecting people's lives.

Using AI to create fake data, researcher may be banned from operations for a limited time

HẠO THIÊN |

Plagiarism, data falsification or the use of artificial intelligence (AI) can cause individuals to be banned from participating in research activities indefinitely.

AI chatbot makes researchers face a crisis of trust

Cát Tiên |

AI chatbots are causing researchers to face a crisis of confidence as nearly 150,000 fake quotes are discovered in many global scientific articles.

Claude expands drug and material research capabilities with AI

Cát Tiên |

Claude's ability to research drugs and materials with AI is expanded through cooperation between Anthropic and scientific AI company SandboxAQ.