AI: Humanity's greatest opportunity or its greatest test?

Artificial intelligence is rapidly moving into a new era, with systems becoming increasingly capable of reasoning, coding, planning and acting with limited human intervention.

As the race to develop more powerful AI accelerates, the debate is shifting from what artificial intelligence can do to a much more consequential question: can humanity remain in control?

The potential benefits are enormous. Advanced AI could help scientists accelerate discoveries, support the development of new medicines, improve productivity and tackle some of the world's most complex problems.

But the same technology could also be used for cyberattacks, disinformation, manipulation, weapons development and other harmful purposes.

The growing capabilities of AI have prompted increasing concern among researchers and technology leaders. The International AI Safety Report 2026 examines both the capabilities of general-purpose AI and the risks associated with increasingly powerful systems, including cybersecurity threats, biological risks, misinformation, economic disruption and the possibility of losing control over highly autonomous AI.

Among the most prominent voices warning about these risks is Geoffrey Hinton, one of the pioneers of modern deep learning.

After leaving Google, Hinton became increasingly vocal about the potential dangers of advanced AI. He has said that while the probability of AI causing human extinction is difficult to calculate, a figure of around 10 percent should not be dismissed. He has also warned that the timeline for serious AI-related risks could be shorter than he once expected.

Concerns are also emerging from within the AI industry itself. In September 2026, researcher Jacob Coxon, who had worked at both OpenAI and Anthropic, publicly criticized the industry's race toward increasingly capable AI systems. Coxon warned that companies could be "gambling with our lives" as they pursue increasingly autonomous and potentially self-improving systems.

The debate gained further attention after Anthropic alignment researcher Evan Hubinger publicly expressed similar concerns.

Hubinger said he personally estimated a greater than 10 percent chance of AI killing all humans within the next decade, while acknowledging that the industry does not yet have a proven solution for aligning superintelligent AI with human interests.

Anthropic CEO Dario Amodei has also called for greater caution, warning that the rapid development of frontier AI could eventually "outrun our ability to understand and control" increasingly advanced systems.

He has advocated measures including independent safety evaluations, greater cooperation between AI companies and stronger international coordination.

The potential dangers extend beyond today's AI systems. Future highly capable systems could potentially identify software vulnerabilities, assist sophisticated cyberattacks or make dangerous biological research easier to conduct.

Researchers are also concerned about systems that could acquire resources, replicate themselves, circumvent restrictions or pursue objectives in ways their creators did not anticipate.

At the same time, governments and technology companies face intense pressure to keep pace with competitors. That competition could create a race in which developing more capable AI becomes a higher priority than ensuring that those systems remain safe and controllable.

Artificial intelligence therefore presents humanity with a historic dilemma. It could become one of the most powerful tools ever created, accelerating scientific progress and transforming economies.

But if its capabilities advance faster than humanity's ability to understand, regulate and control them, the same technology could create risks unlike anything the world has faced before.