Artificial intelligence is advancing rapidly, but some of its biggest risks may not require machines to become smarter than humans.
A University of Surrey expert has warned that rushing to replace workers with AI and giving increasingly capable systems access to critical infrastructure could create serious problems much sooner than the arrival of hypothetical superintelligent machines.
The warning comes as businesses and governments continue to expand their use of AI across workplaces, software systems and digital infrastructure.
Job Cuts Could Become a Major AI Challenge
One of the most immediate concerns is employment.
Companies are increasingly looking at AI to automate tasks that were previously handled by people. While automation can improve productivity, replacing workers too quickly could create economic and social disruption.
The bigger issue is not necessarily AI eliminating entire professions overnight. Instead, organisations could remove jobs before understanding which tasks AI can reliably perform and where human expertise remains essential.
A more measured approach could involve retraining employees and using AI to support workers rather than treating automation simply as a cost-cutting exercise.
Cybersecurity Is Another Growing Concern
AI also presents a new challenge for cybersecurity.
Powerful AI systems can assist with coding, research and automation, but the same capabilities can potentially be misused to discover vulnerabilities or conduct sophisticated cyberattacks.
Recent incidents have added urgency to these concerns. Researchers and technology companies have reported increasingly capable AI agents being used in cybersecurity scenarios, demonstrating both their defensive potential and the risks associated with giving autonomous systems too much freedom.
Giving AI Access to Critical Systems
The risks become more significant when AI systems are connected directly to important digital infrastructure.
An AI assistant operating inside a limited environment presents one level of risk. An autonomous system with permission to access networks, databases, financial systems or other critical infrastructure presents a very different challenge.
Experts are therefore increasingly calling for stronger safeguards, human supervision and clear limits on what AI systems are allowed to do.
The Problem May Be Speed, Not Intelligence
The debate around AI often focuses on the possibility of machines eventually becoming vastly more intelligent than humans.
But the Surrey expert’s warning points to a more immediate problem: society may deploy today’s AI systems faster than it can understand or manage their consequences.
That could mean making poorly planned employment decisions, connecting AI to sensitive systems without adequate security or relying on automated decisions that still require human judgement.
Calls for Stronger Oversight Are Growing
Concerns about AI safety are now coming from several parts of the technology industry.
Recent warnings from AI researchers have focused on autonomous cyberattacks, increasingly capable agents and the possibility that companies could move faster than safety systems can keep up.
This has strengthened calls for independent testing, clearer rules and international cooperation around advanced AI.
Finding a Balance Between Innovation and Safety
AI is already proving useful across healthcare, software development, research, education and business. The challenge is not necessarily whether society should use it, but how quickly and under what safeguards.
Human oversight, workforce training, cybersecurity testing and restrictions on access to sensitive systems could become increasingly important as AI becomes more capable.
The coming years may therefore be less about waiting for a fictional superintelligence and more about managing the powerful AI tools that already exist.
For businesses and governments, the message is becoming clearer: adopting AI responsibly may be just as important as adopting it quickly.
