The story the media tells about artificial intelligence and the story the public is living are at odds, according to new research from CARMA.
“Hype, Hope or Harm? The Global Story of AI,” a study examining coverage across 500 global media outlets and attitudes among 6,300 people in 19 markets, reveals a gap between media narratives and public concern.
In nearly 70% of markets studied, media coverage signals higher trust in AI than the public actually feels. Nearly six in ten media stories about AI in 2025 recorded a positive sentiment, celebrating breakthroughs in generative assistants, search capabilities, and operational efficiency. “Coverage peaks around product launches and funding rounds, then quietly recedes, a rhythm that suggests excitement without staying power,” said CARMA.
However, among the public, the conversation is different. More than half of people surveyed worry about cybercrime, hacking, and fraud enabled by AI. Nearly as many fear misinformation, deepfakes, and election interference.
The public isn’t asking whether AI will make businesses more efficient. More than half of people surveyed worry about cybercrime, hacking and fraud enabled by AI.
Media coverage, by contrast, lingers on more abstract concerns such as loss of human oversight, a risk that registers with just over a quarter of the public.
Who’s Talking?
Chief executives account for 43% of AI commentary in the media. Journalists contribute just 2% of commentary on a technology reshaping their own profession, said the report.
The result is a narrative weighted towards business benefits and transformation, told by those with the most to gain from AI’s adoption.
“Academics and regulators add credibility, but with less visibility. NGOs and unions, more skeptical voices, struggle to break through,” said CARMA.
Significant Geographic and Demographic Divides
Geographic and demographic divides are significant. In India, China, Saudi Arabia, and Brazil, AI is embraced as an engine of progress. In Japan, France, Italy, and Canada, it provokes anxiety. The US, UK, and Australia occupy a middle ground, cautiously positive but alert to risk. European media gravitates towards risk narratives. In the Americas, Asia-Pacific, the Middle East, and Africa, AI’s impact is broadly framed as manageable.
European media gravitates towards risk narratives. In the Americas, Asia-Pacific, the Middle East and Africa, AI’s impact is broadly framed as manageable.
Age matters too. Among those under 20 and over 60, fear outweighs excitement. They see inevitability but little agency.
According to the report, responsible AI positioning lacks differentiation. Companies trying to lead on safety and ethics faced a challenge last year to make their messaging resonate. Even organizations strongly positioned around responsible AI showed limited differentiation in coverage. The market lacked a shared language for what responsible AI leadership looks like.
The report adds that things may be changing. In 2026, companies including Anthropic have stepped up their focus on safety and transparency, signaling a potential shift in how the industry talks about responsibility. Whether this breaks through the noise remains to be seen.
The first movers in responsible AI positioning haven’t broken through yet, which means the groundwork is still being laid.
CARMA points to this as both a challenge and an opportunity. “The first movers in responsible AI positioning haven’t broken through yet, which means the groundwork is still being laid,” said CARMA.
For organizations navigating AI’s reputational landscape, CARMA’s research points to a path forward: address the gap head-on, speak to oversight and control rather than efficiency alone, prioritize public concerns around fraud and misinformation over the risks that dominate conference panels, elevate more voices beyond the CEO, and tailor messaging to each market’s maturity and demographic makeup.
“The story of AI is still being written,” said CARMA. “But the gap between how it’s told in the media and how it’s experienced by the public suggests the next chapter needs different authors, and a more honest plot.
Study Highlights
Carma provided the following summary of highlights from the report.
Optimism in AI coverage is strong but hype/event-driven
57% of AI media coverage in 2025 was positive. However, positive sentiment tends to spike around product launches, conferences and major investment announcements, before falling back. This suggests optimism is often tied to news events rather than sustained narratives about real-world deployment.
AI narratives vary significantly by market
High-growth markets such as India, China, Saudi Arabia and Brazil show strong enthusiasm for AI and respond well to transformation narratives. In contrast, Western Europe, Canada and Japan show greater caution, with stronger emphasis on governance and risk.
Media coverage does not fully reflect public concerns
In most markets, media coverage presents AI more positively than public opinion. Audiences are most concerned about cybercrime, fraud, misinformation, deepfakes and election interference, which receive less emphasis in coverage than issues such as loss of human oversight.
CEOs lead the AI conversation
Chief executives account for 43% of AI commentary in media, making them the most prominent voice in shaping narratives around AI’s business impact. While this gives companies narrative influence, it also concentrates scrutiny when coverage becomes more critical.
Responsible AI positioning is not yet clearly differentiated
Companies emphasizing safety and ethics in their messaging do not consistently stand out in media coverage. Even firms strongly positioned around responsible AI show limited differentiation in visibility or sentiment.
The full report is available here.
