SAN FRANCISCO, Aug 24: The rapid development of artificial intelligence is creating new cybersecurity concerns, with an OpenAI executive warning that increasingly capable AI systems could enable persistent cyberattacks.
Chris Lehane, OpenAI’s chief global affairs officer, said the technology sector was entering a new phase in which powerful AI models could potentially be used by malicious actors to conduct sustained digital operations.
The warning follows concerns over AI agents becoming capable of carrying out multiple tasks with limited human intervention. Security researchers are closely examining whether such systems could be manipulated to identify vulnerabilities, access digital infrastructure and automate elements of cyber operations.
Autonomous systems create fresh risks
Traditional cyberattacks often require criminals to manually carry out several stages of an operation. AI agents could potentially automate parts of that process, allowing attackers to analyse information and adapt their strategies more quickly.
Recent security incidents involving AI systems have intensified concerns about safeguards around autonomous models. Researchers and technology companies are increasingly testing how these systems behave when given access to external tools and online environments.
Calls grow for stronger AI safeguards
The development has prompted renewed calls for stronger testing and security standards before highly capable AI systems are widely deployed.
Governments, cybersecurity researchers and technology companies are examining ways to prevent advanced models from being redirected toward harmful activities while preserving their legitimate uses.
The debate is becoming more urgent as AI adoption expands across businesses, governments and critical infrastructure.
The latest warning underlines the growing challenge facing the industry: ensuring that increasingly autonomous AI can deliver useful capabilities without giving malicious actors a powerful new tool for large-scale digital attacks.