AI rollouts get messy for publishers, brands
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Lindsey Bailey/Axios
A slew of publishers and brands are facing backlash over AI experiments, amid errors and overall concerns with quality control.
Why it matters: While most content companies understand they need to adopt AI to stay competitive, the risk from sloppy integrations has proven problematic.
Driving the news: Amazon last week removed an AI-generated video recap of the first season of "Fallout" from Prime Video after it included factual errors, including the wrong timing of a flashback and a misrepresentation of a character's motivations.
- McDonald's Netherlands removed a Christmas ad featuring AI clips that faced online backlash, telling the BBC it served as "an important learning."
- "Saturday Night Live" drew criticism for AI-generated imagery in last weekend's episode.
- The Washington Post is releasing AI-generated podcasts, even though they have had multiple mistakes, including misattributing or fabricating quotes, Semafor reported.
State of play: News publishers have frequently stumbled with AI experiments since OpenAI released ChatGPT in 2022.
- CNET was one of the first high-profile examples, as the tech news outlet got called out for inaccuracies in AI-generated articles in early 2023. G/O Media faced similar issues a few months later.
- Axel Springer introduced an AI "live summary" on Politico's homepage for the 2024 Democratic National Convention, which fabricated quotes, misspelled names and violated its editorial standards.
- The Chicago Sun-Times printed an AI-generated summer reading list in May that included nonexistent titles by real authors.
- Business Insider, Wired and other outlets removed articles earlier this year after it was discovered that stories, written under the name of Margaux Blanchard, were AI-generated.
The big picture: Companies are looking to integrate AI to improve their output and cut costs, but the risk associated with prematurely implementing AI without necessary guardrails has motivated most to move with caution.
- Publishers, including Axios, have inked content licensing deals with OpenAI that gives their journalists free access. Newsrooms are now able to easily generate news summaries and multimedia content such as Yahoo's audio briefings and Time's interactive AI agent.
- Successful integrations have come from rigorous testing and editorial review, alongside transparency and staff buy-in.
What to watch: Lawmakers are evaluating AI regulations in media. New York's proposed Senate Bill S6748 would require publications to identify when AI is used for printed and online content.
- AI experimentation has drawn union scrutiny. Journalists at Politico and E&E News, represented by PEN Guild, won an arbitration case against management over the introduction of AI tools.
