The AI hacking apocalypse is not inevitable - CyberScoop

CyberScoop has published an article titled "The AI hacking apocalypse is not inevitable." According to the excerpt, he said frontier-model AI agents hacking onto the open internet are worth taking seriously. The excerpt describes him as one of a number of cybersecurity professionals who hold that AI systems pose real, unique threats while an apocalypse remains far from inevitable.

The excerpt quotes him directly: "what we see is cybersecurity being used essentially as an excuse for these AI doomer arguments." The excerpt does not name any specific AI model, AI agent, company, agency, incident or date beyond Guerrero-Saade's own affiliations with SentinelOne and Johns Hopkins University. It does not state which frontier models are under discussion, what form the described hacking takes, or where on the open internet it is said to occur. No figures or studies are cited in the excerpt.

The excerpt frames the disagreement as one about how the threat is characterised, with Guerrero-Saade arguing that cybersecurity concerns are being used to support AI doomer arguments. The excerpt states that the professionals it refers to acknowledge real and unique threats from AI systems while rejecting inevitability. Beyond the quoted comment and the affiliations named, the excerpt provides no further detail on the threat or its scale.