Social media is exploding with deepfakes. Sophisticated new tools can clone real voices to create AI audio that can fool listeners and spread disinformation. 

TikTok has blown up with voices added to AI-generated videos. ChatGPT creates humanlike writing and services like Midjourney generate instant images and art. 

In the digital age, the lines between what’s real and what’s not are becoming increasingly blurred. 

The rise of artificial intelligence (AI) audio technology has opened up new frontiers in creating convincing fake audio content. As this technology gains traction, companies like ElevenLabs have unveiled a suite of tools designed to harness AI voices for various applications.

 These developments, however, are not without their controversies, as they have given rise to a new battleground for online misinformation. Chatbots, powered by AI-generated audio, have become a potent force in amplifying disinformation, posing a significant threat as we approach the 2024 election.

The Emergence of Chatbots Powered by AI Audio

At the forefront of this audio revolution is a new generation of chatbots that rely on AI-powered voices. 

The New York Times describes 4 chatbots as potential game-changers that could disrupt the established order of the internet. 

1. ChatGPT: Developed by the research lab OpenAI, ChatGPT has garnered significant attention since its launch in November.

This AI language model boasts a remarkable ability to tackle complex questions, generate poetry, write code, plan vacations, translate languages, and now, even respond to images.

GPT-4, introduced in mid-March, takes this technology to a new level, demonstrating a high level of versatility and sophistication.

2. Bing: Following closely on the heels of ChatGPT, Microsoft, OpenAI's primary investor and partner, integrated a similar chatbot into its Bing internet search engine. 

This chatbot can engage in conversational texting on a wide range of topics. However, its occasional inaccuracies, misleading responses, and quirky behavior have sparked considerable interest, for better or worse.

3. Bard: Google's Bard, released in March to a select group of users in the United States and Britain, was initially designed as a creative tool to draft emails and poems.

However, it has since demonstrated its capacity to generate ideas, write blog posts, and provide factual or opinion-based answers to questions.

4. Ernie: Baidu, the Chinese-based tech giant, unveiled "Ernie" in March, as a competitor to ChatGPT. Ernie represents a significant player in the AI audio arena.

Don’t Miss Out on the Ring of Fire Eclipse: How to Safely View This Cosmic Event - USA Herald

NASA’s Psyche Mission Marks a Milestone in Space Exploration SpaceX Falcon Heavy Launch - USA Herald

Disinformation Watchdogs on High Alert

As chatbots powered by AI audio gain prominence, disinformation watchdogs have observed a worrying surge in the use of AI-generated voices in videos and other content. 

These novel tools have been embraced by content producers and purveyors of misinformation, particularly on platforms like TikTok. 

One company at the forefront of this battle is NewsGuard, a company specializing in monitoring online misinformation.

In a report released in September, NewsGuard identified 17 TikTok accounts that were actively disseminating baseless claims using AI-generated audio. These accounts primarily focused on celebrity rumors, often accompanied by narration from AI voices.

These accounts collectively amassed hundreds of millions of views and received comments suggesting that some viewers believed these unfounded claims.

Jack Brewster, the enterprise editor at NewsGuard, commented on this phenomenon: "It's a way for these accounts to gain a foothold, to gain a following that can draw engagement from a wide audience. Once they have the credibility of having a large following, they can dip their toe into more conspiratorial content."

While TikTok does require labels disclosing AI-generated content as fake, NewsGuard's flagged videos did not display such labels.

Public perception

 TikTok, in response, removed or stopped recommending several of these accounts and videos, citing violations of policies related to impersonation of news organizations and spreading harmful misinformation. A particular video featuring an AI-generated voice mimicking former President Obama was also removed for violating TikTok's synthetic media policy, as it presented highly realistic content without the appropriate labeling.

The challenges posed by AI-generated audio extend beyond TikTok. Other major platforms like YouTube, Instagram, and Facebook have also witnessed the proliferation of AI-generated content. These platforms permit AI-generated content featuring public figures, including newscasters, as long as it does not disseminate misinformation. 

Parody videos, which depict AI-generated conversations between politicians, celebrities, and even deceased individuals, have become widespread as these AI tools gain popularity.

The impact of AI-generated audio and video is undeniable. Claire Leibowitz, head of A.I. and media integrity at the Partnership on A.I., emphasized, "We do know audio and video are perhaps more sticky in our memories than text." These technologies have the potential to profoundly influence public perception.