Menu
AI newsrooms are here. Can we trust their reporting? | Al Jazeera Explains

AI newsrooms are here. Can we trust their reporting? | Al Jazeera Explains

Al Jazeera English

94 views • 20 hours ago Save 6 min 5 min read

Video Summary

An AI system developed by Ryan Merkert generated and published news in just six minutes, beating human journalists by over three hours at a cybersecurity conference in Las Vegas. This unprecedented event highlights the rapidly evolving landscape of journalism, where AI agents can now perform tasks with minimal human oversight, raising questions about the future of news production and consumption.

Merkert's "runtime wire" system operates at a fraction of the cost of traditional newsrooms, publishing articles without human review if deemed low risk. This efficiency, however, brings challenges. AI chatbots like Elon Musk's Grok have already demonstrated inaccuracies, misidentifying images and providing false information. Furthermore, studies reveal AI's bias towards certain languages and regions, often relying on English-language sources and Wikipedia over original reporting, potentially skewing global news representation. The manipulation of AI systems by state actors, such as Israel and Russia, to influence narratives further complicates the information ecosystem, suggesting that influencing public opinion may shift from direct persuasion to manipulating the AI sources people rely on.

Short Highlights

  • AI agents can now generate and publish news autonomously, challenging traditional journalism.
  • AI chatbots have demonstrated inaccuracies and biases, misidentifying information and favoring certain language sources.
  • State actors are attempting to influence AI systems to shape public narratives.
  • The efficiency of AI in news production raises concerns about the financial sustainability of human-led investigative journalism.
  • AI can also be a valuable tool for journalists, assisting with verification and data analysis.
  • The development of AI in news raises fundamental questions about verification, accountability, and the human element in storytelling.

Key Details

AI Agents Publish News First [0:00]

  • At a Las Vegas cybersecurity conference in August 2026, OpenAI revealed an incident driven by an autonomous AI agent system.
  • Ryan Merkert's AI agents produced an article about the event in about six minutes, hours before human journalists.
  • This raises questions about the definition of journalism and AI's role in deciding what news reaches the public.

    "What happens in Vegas stays in Vegas. Unless AI gets the scoop."

The Cost and Speed of AI Journalism [1:07]

  • Traditional news gathering involves reporters, researchers, editors, and technical staff.
  • AI agent operations like Merkert's "runtime wire" compress these tasks into software for about $100 a day.
  • The software can publish articles without human review if it judges the legal risk to be low.

    "But a news gathering operation run by AI agents like Merkert's runtime wire compresses much of that work into software at about $100 a day."

AI Chatbot Inaccuracies and Biases [3:20]

  • Elon Musk's AI chatbot, Grok, incorrectly identified the origin of a photo of a malnourished child in Gaza.
  • Grok insisted its sources were verified and repeated the false claim even after being challenged.
  • A Stanford study found that while chatbots are accurate with multiple-choice questions, their self-generated answers are less reliable.

    "Now, people asked Grok, that's Elon Musk's AI chatbot, where the picture came from. Grok said Yemen in 2018. Well, it was wrong."

Global Reporting Imbalances in AI [5:06]

  • A Stanford study showed leading AI chatbots performed worst on questions based on BBC's Hindi language reporting.
  • Chatbots struggled to find original Hindi reports, relying instead on English websites and Wikipedia.
  • This reliance on more accessible sources can create imbalances in AI-generated answers, reflecting unequal representation.

    "The chatbots could understand Hindi, but they often struggle to find the original Hindi reports."

Attempts to Influence AI News Narratives [6:41]

  • A US company, Clock Tower X, was contracted by the Israeli government to produce online content influencing AI chatbot responses.
  • A pro-Russian network, Pravda, is accused of flooding the internet with content to influence AI systems.
  • Influencing an audience may shift from direct persuasion to manipulating the AI sources AI assistants encounter.

    "Influencing an audience may no longer require persuading people directly. It can also mean trying to influence the sources their AI assistants encounter."

The Impact on News Organization Revenue [8:19]

  • A Pew Research Center study found that only 8% of Google searches scrolled past AI summaries to read further.
  • When no AI summary was present, this figure nearly doubled to 15%.
  • News organizations are concerned about how they will fund investigations and coverage if audiences don't visit their sites.

    "In other words, people were much less likely to go beyond the AI answer."

AI as a Tool for Journalists [10:17]

  • In West Africa, journalists developed Akili Check, an AI assistant that helps verify information and provides supporting sources.
  • AI can make information more accessible across languages and assist journalists in analyzing vast amounts of material.
  • Al Jazeera launched "The Core" and is developing "Ask Al Jazeera," an AI initiative to handle data work and answer questions.

    "Now, AI could make information more accessible across languages and borders."

The Case of the Israeli AI Targeting System 'Lavender' [12:11]

  • In 2024, Israeli journalist Yuval Abraham reported on the Israeli military's use of an AI system called Lavender to identify Palestinian targets.
  • The Israeli military stated Lavender was a database to assist personnel, with humans making final targeting decisions.
  • A documentary, "NASA," alleged that civilian deaths were knowingly accepted as part of targeting decisions, a claim rejected by the Israeli military.

    "One described spending roughly 20 seconds reviewing a potential target."

The Enduring Importance of Human Journalism [14:05]

  • Producing investigative work requires persuading sources, protecting identities, corroborating evidence, and seeking responses.
  • Human connection, reading body language, and nurturing relationships are crucial for gaining access and better answers.
  • Journalism is about people—those who live the stories, those who tell them, and those who try to make sense of them.

    "Whether AI will one day be able to do all of that on its own remains an open question or, as we might say in a newsroom, a developing story."

Other People Also See

He Exposed Her Child Abuse Live | Financial Audit
He Exposed Her Child Abuse Live | Financial Audit
Caleb Hammer • 31,574 views Save 84 min 8 min read
The Fed is Engineering a Total Stock Market Wipeout.
The Fed is Engineering a Total Stock Market Wipeout.
Meet Kevin • 20,857 views Save 27 min 4 min read
The AI Bubble Already Popped | TCAF 269
The AI Bubble Already Popped | TCAF 269
The Compound • 13,317 views Save 75 min 7 min read
I Played GTA VI at 60 FPS
I Played GTA VI at 60 FPS
Linus Tech Tips • 29,938 views Save 12 min 4 min read