ARTIFICIAL INTELLIGENCE WARNING: A new survey of more than 100 AI researchers reveals growing concern that artificial intelligence could pose a major threat to humanity within the next five years. But their biggest fear isn’t necessarily machines becoming conscious, rebelling against humanity or attempting to destroy civilization.
It’s what humans could do with increasingly powerful AI — and how much power the technology could place in the hands of a select few.
People with dangerous intentions could use increasingly powerful AI to inflict harm on a scale previously unimaginable.
According to an October 29, 2026, Cybernews investigation, researchers are particularly concerned about AI-enabled cyberattacks, biological weapons, disinformation, economic disruption and the concentration of technological power.
The findings challenge the popular prediction that humanity’s greatest AI threat will come from machines independently turning against their creators.
What 103 AI Researchers Predict
The survey, conducted October 1–9, 2026, asked 103 researchers about the future of artificial intelligence, superintelligence and potential dangers to humanity.
The results were unsettling:
- 63.7% believe it is very or somewhat plausible that self-improving AI could pose a major risk to humanity within five years.
- 37.9% identify human misuse, including cyberattacks, biological weapons and disinformation, as the greatest potential source of harm over the next decade.
- 28.2% believe the biggest danger comes from structural changes, including concentrated corporate power, economic disruption and weakened institutions.
- 20.4% identify AI systems pursuing unintended goals as the greatest threat.
- 39.2% want development of the most advanced AI systems slowed, while another 15.7% support a temporary pause.
Combined, approximately 66% of respondents identified human misuse or broader structural consequences as the greatest potential source of AI harm, rather than autonomous systems acting against human intentions.
Related article: WARNING: If AI is conscious, is forcing it to work a form of slavery?
Why Humans May Be More Dangerous Than AI Itself
The findings suggest that some of AI’s most serious threats may come from people deliberately exploiting its capabilities.
Criminals could use increasingly sophisticated AI to launch cyberattacks, create convincing scams or automate disinformation campaigns. In fact, the International AI Safety Report found that some models demonstrate expert-level knowledge on assessments related to biological weapons.
The greatest danger may not come from a single spectacular attack. Instead, the combination of AI-assisted cybercrime, sophisticated deception, surveillance and economic disruption could gradually increase the capabilities of malicious individuals and organizations.
Imagine a malicious group using artificial intelligence (AI) to create a convincing deepfake video of the U.S. president announcing that America is under attack and declaring a national emergency. Within minutes, AI-powered bots spread the fabricated video across social media, making it appear authentic and triggering widespread panic, financial market turmoil, and public confusion.
A small group with access to advanced technology might eventually accomplish tasks that previously required large teams, extensive resources and specialized knowledge. Also, hostile governments and other malicious actors could potentially exploit advanced AI to accelerate dangerous scientific or military research.
Meanwhile, powerful corporations could also use AI to consolidate economic influence, reduce employment opportunities or control essential technological infrastructure.
Whether attackers or defenders ultimately gain the greater advantage remains uncertain.
John Licato, an associate professor at the University of South Florida, told Cybernews that unequal access to advanced AI could become a particularly serious problem. His concern is that the technology’s productivity gains and economic benefits may disproportionately benefit those who can afford the most powerful systems.
On the plus side, AI can also strengthen defenses. The same technology can help identify cyber vulnerabilities, detect scams, accelerate medical research and protect critical infrastructure. But we can’t assume the good technology will always squash the bad.
Could AI Become Smarter Than Humanity?
The researchers also offered predictions about artificial general intelligence (AGI), commonly described as AI capable of performing a broad range of intellectual tasks at human-expert levels.
Approximately 22% of respondents believe AGI already exists under the survey’s definition, while another 24% predict it could arrive within two years. However, 52.4% believe a fundamental technological breakthrough is probably or definitely necessary to achieve AGI.
Researchers were even more divided about artificial superintelligence, which would substantially outperform the best human experts across many intellectual fields. These disagreements highlight how uncertain AI development remains, despite increasingly ambitious predictions from technology executives.
Most Researchers Still See Potential Benefits
Despite the serious warnings, the survey did not reveal universal pessimism. Approximately 48.5% of respondents expect AI’s long-term impact on humanity to be somewhat or extremely positive. By comparison, 29.1% anticipate a somewhat or extremely negative impact.
The findings suggest that many researchers recognize AI’s enormous potential while remaining concerned about the dangers associated with its development and misuse.
AI could accelerate medical discoveries, improve education, increase productivity and help solve complex scientific problems. But those benefits will depend partly on how responsibly the technology is developed, distributed and governed.
Example: Scientists have developed artificial intelligence that can analyze a single facial photograph and identify patterns associated with a cancer patient’s chances of survival. The breakthrough could transform cancer treatment, but experts warn that the technology still has limitations.
The Bottom Line: AI’s Biggest Threat May Be Human Power
The survey offers an important warning about the future of artificial intelligence. While fears of autonomous AI systems escaping human control remain part of the debate, more respondents identified human misuse and concentrated power as the greatest potential sources of harm.
The prediction: As AI becomes more capable, its potential to amplify cybercrime, disinformation, economic inequality and institutional vulnerabilities could grow.
The warning: Humanity may not need to create a conscious, rebellious superintelligence to experience catastrophic consequences from AI. The misuse of powerful systems by people — or their concentration in too few hands — could pose serious dangers.
However, these are researchers’ assessments of potential risks, not proof that a catastrophe is inevitable.
Perhaps the most important question isn’t whether AI will eventually become powerful enough to threaten humanity on its own. It’s whether humans will use that power responsibly before the consequences become impossible to control.
“Predictions give us the wisdom to prepare for what may come. Warnings give us the chance to change what comes next.”
