Why blocking AI models won’t stop the cyber threats they create

The piece states that new AI models now possess skills equivalent to the best human hackers, making them highly capable tools for cyber attacks. The author points out that measures to control access serve only as temporary barriers.

The article contends that determined adversaries will find ways to access or replicate such models, rendering restrictions ineffective. It suggests that the cybersecurity community must adopt strategies that go beyond simple blocking, such as improving defensive AI and system resilience. The op-ed emphasises the need to recognise that capable AI models are here to stay and will continuously challenge security efforts.

The piece does not name specific models or incidents, but its arguments align with ongoing debates about the risks of advanced AI. The author calls for proactive and adaptive approaches to cyber defence, rather than relying on the illusion of control through access restrictions. The op-ed warns that the pace of AI development is outstripping efforts to contain its malicious uses.