UN accused of using AI to write press releases

The United Nations (UN) is among organizations increasingly relying on artificial intelligence (AI) systems for written communication. A study has found that 13% of its press releases have patterns resembling AI.

A new study published in the Cell Press journal Patterns reveals that 17% of analyzed corporate and governmental written content, including job posts and press releases, may have been written using AI tools.

Consumers also appear to frequently use large language models (LLMs) for written communication – the study suggests that about 18% of complaints submitted to the Consumer Financial Protection Bureau between 2022 and 2024 were likely AI-assisted.

That’s no surprise, given that more than half (52%) of American adults report using LLMs, mostly for personal, informal learning purposes, but also increasingly for work.

The researchers found that nearly a quarter of corporate news releases published in three major news release platforms were AI-generated, with science and technology releases having the highest rate of AI use.

Small companies turn to AI for job postings

Large companies were less likely to use AI to write job postings on LinkedIn. In contrast, about 10% of vacancy postings from smaller firms were likely written with AI assistance.

The findings come as job seekers are increasingly frustrated with AI's use in the hiring process. While using LLMs to write job postings is considered one of the most comfortable AI applications, many want companies to be more upfront about using AI.

The analysis of UN press releases written in English reveals a significant increase in AI-assisted writing, from 3% in early 2023 to more than 13% by late 2024.

The UN has previously warned that AI threatens one in four jobs, with women in high-income countries being especially vulnerable to AI-driven task automation.

In the United States, communications and PR are female-dominated fields. According to 2024 data, women compose 67.8% of public relations specialists and 70.1% of PR and fundraising managers.

Overall, the researchers found that the portion of content flagged as written mainly by AI increased sharply from 1.5% before the release of ChatGPT in November 2022 to more than 15% by August 2023.

Growth slowed down later, and the AI adoption rate was about 17% by August 2024.

AI detection tools are often unreliable

The researchers used an AI detection program that they had previously developed to investigate the adoption patterns of these AI tools. However, they admitted that the tool cannot accurately differentiate texts that are heavily edited by humans.

This reflects the previous findings suggesting major drawbacks of AI-text detection tools.

Have thoughts about this topic? Others do, too. Join them in the discussion.

A 2023 study concluded that such tools are "neither accurate nor reliable," as they all scored below 80% in accuracy tests. Many tools classified AI-generated text as human-written rather than detecting AI.

Their performance further worsened when using techniques like manual editing or machine paraphrasing, or when they analyzed texts translated from other languages.

Another study suggested that AI text detection tools' already low accuracy rates may drop to 17.4% when faced with manipulated content.

Corresponding author James Zou of Stanford University says the AI detection tool works best with a large collection of text, but would be unable to pinpoint whether a single article used AI.

Zou says in a press release, "Like all new technologies, it's difficult to say if these AI models are simply 'good' or 'bad.' They still make mistakes, so if people completely outsource their job to these tools and don't bother to check the accuracy, that could lead to errors in their writing."