CARMA’s latest Perceptions of AI report reveals how growing concerns over safety, accountability, and human control are reshaping global AI narratives, and what this means for business leaders and communicators.
Artificial intelligence continues to transform industries, attract investment and drive innovation. But the global conversation around it is changing.
For the first time since CARMA began tracking emotional signals in AI media coverage, fear and concern have overtaken excitement. At the same time, questions about AI safety, corporate responsibility and the consequences of rapid technological development are gaining prominence.
CARMA’s third edition of Perceptions of AI: Progress Under Pressure examines 3,481 articles from 500 leading media outlets across 44 markets, covering July to September 2026.
The findings reveal a critical shift: AI is still widely portrayed as an opportunity, but technological progress alone is no longer enough to sustain positive perceptions.
How has global media sentiment towards AI changed in 2026?
Global media coverage is becoming more cautious as enthusiasm for AI increasingly competes with scrutiny of its risks.
In the first half of 2026, 47% of AI media coverage was positive. By Q3, that figure had fallen to 32%, while balanced coverage increased from 12% to 32%.
This shift reflects a more complex conversation. AI’s potential to improve productivity and accelerate innovation remains significant, but media attention is increasingly focused on whether its development is progressing faster than the safeguards designed to govern it.
By September, the narrative that « AI needs guardrails » had overtaken the previously dominant « AI as a productivity miracle » narrative.
For organisations communicating about AI, demonstrating innovation is no longer enough. Credibility increasingly depends on explaining how technology is being developed, deployed and managed responsibly.
Why is fear overtaking excitement about AI?
In September 2026, fear and concern accounted for 38% of emotional signals in AI media coverage, compared with 32% for excitement.
This marked the first time fear and concern had overtaken excitement since CARMA began tracking these signals.
The increase was driven by media reporting on autonomous AI incidents, cybersecurity concerns and warnings from industry insiders about the pace of AI development and the limits of human oversight.
However, the emotional response is far from uniform across markets.
European media displayed particularly strong signals of fear and concern, including Germany (55%), the UK (48%) and France (45%). In contrast, excitement remained the dominant emotional signal in China and South Korea.
These differences highlight an important challenge for global communicators: AI does not have one universal reputation.
Messaging that resonates in markets focused on technological opportunity may require a different approach in markets where safety, governance and societal consequences dominate the conversation.
Who is shaping the global AI conversation?
CEOs remain the most prominent voices in AI media coverage, accounting for 42% of leading commentator voices in Q3 2026, an increase of 12 percentage points compared with H1.
Leaders including Sam Altman, Dario Amodei and Elon Musk continued to dominate the conversation. Yet their visibility was increasingly associated with debates about governance, existential risk and the pace of AI development rather than innovation alone.
There is also a significant distinction between visibility and credibility.
While CEOs remain among AI’s strongest advocates in media coverage, CARMA’s audience research identifies them as one of the least-trusted commentator groups.
This presents a reputational challenge for organisations relying heavily on executive communications to shape perceptions of AI.
As scrutiny increases, leadership statements need to be supported by tangible evidence, transparent practices and credible independent voices.
Is AI still seen as an opportunity?
Yes. Despite growing concern, opportunities continue to account for the majority of global AI media discussions about benefits and risks.
However, the gap is narrowing.
In 2025, opportunity narratives represented 72% of coverage, compared with 28% for risk and harm. By Q3 2026, the balance had shifted to 59% opportunity and 41% risk and harm.
The findings do not suggest that media coverage is rejecting AI’s potential. Instead, expectations are becoming more demanding.
Productivity, workplace efficiency and economic development remain important benefits, but increasingly, discussions focus on demonstrated outcomes rather than transformational promises.
At the same time, more severe risk narratives are emerging, particularly around autonomous AI systems operating beyond meaningful human oversight and AI-enabled cybercrime.
For businesses, the challenge is shifting from communicating what AI could achieve to demonstrating what it has achieved, while acknowledging and addressing the risks involved.
Why are AI data centres becoming a reputational concern?
The rapid expansion of AI infrastructure has introduced another dimension to the global reputation debate: the impact on local communities.
Data centres promise investment, economic development and technological competitiveness. However, concerns about electricity consumption, water resources, household costs and environmental impact are generating growing public resistance.
CARMA’s analysis found that 46% of global media coverage discussing AI’s impact on local communities was balanced, while 31% was negative and just 21% positive.
The research highlights a clear distinction between positive and negative infrastructure narratives.
Positive stories gained traction when organisations demonstrated tangible benefits such as employment, tax revenues and investment in local services. Negative coverage intensified when economic promises were questioned or communities felt excluded from decisions affecting them.
For organisations investing in AI infrastructure, securing trust requires more than communicating the scale of investment. It requires demonstrating who benefits, addressing local concerns and ensuring communities have a credible voice.
What do these findings mean for PR and communications professionals?
AI’s evolving reputation creates new challenges for organisations communicating innovation, transformation and technological progress.
Based on the research, five priorities stand out:
1. Lead with evidence, not promises. Demonstrate measurable outcomes, practical applications and tangible benefits rather than relying on broad claims about transformation.
2. Make AI safety part of reputation management. Communicate safeguards, human oversight and accountability before technical incidents escalate into reputational crises.
3. Build credibility beyond executive visibility. Support leadership messaging with independent expertise, transparent evaluation and verifiable evidence.
4. Adapt communications to regional expectations. Account for differences in how markets perceive AI’s opportunities, risks and societal impact.
5. Monitor emotional signals alongside media sentiment. Understand not only whether coverage is positive or negative, but which developments are generating fear, anger, trust or excitement.
As the global AI conversation becomes increasingly contested, organisations will need to demonstrate that innovation and responsibility can advance together.
Frequently asked questions about AI perceptions in 2026
Is public trust in AI declining?
CARMA’s Q3 2026 media analysis shows growing scrutiny of AI, but media sentiment should not be treated as a direct measure of public trust. The findings indicate that credibility increasingly depends on transparent communication, demonstrable safeguards and responsible AI deployment.
What are the biggest concerns surrounding AI in 2026?
The most prominent emerging risk narratives in CARMA’s Q3 analysis involve AI systems operating beyond meaningful human oversight and AI-enabled cybercrime. Concerns about job displacement, misinformation, privacy and accountability also remain important.
How can organisations build trust when communicating about AI?
Organisations can strengthen credibility by demonstrating measurable benefits, communicating safeguards clearly, involving independent experts and addressing the specific concerns of their audiences. Responsible action, supported by verifiable evidence, provides a stronger foundation than broad technological promises.
The next phase of AI reputation
The global AI narrative is moving beyond the excitement of technological breakthroughs towards a more demanding discussion about safety, responsibility and real-world consequences.
The opportunity remains significant. But as scrutiny increases, the organisations best positioned to build lasting credibility will be those that can demonstrate not only what AI makes possible, but how its benefits are delivered responsibly.
Explore the complete findings in CARMA’s Perceptions of AI 3: Progress Under Pressure report.
