How much will cybersecurity costs rise?

A new estimate from governance.ai suggests the UK could face an additional $100 billion in annual cybercrime costs, a 20% increase, as AI-related threats rise. The figure is one of several projections cited in a discussion about quantifying cybersecurity expenses, with others from superforecasters and AI agent-based analyses. The author calls for more rigorous, GDP-share-based estimates to temper emotional responses to AI safety concerns.

How much will cybersecurity costs rise? Here is one partial estimate of the higher cyber costs https://x.com/sj manning/status/2086804037286896011 . I am not sure how many incidents they have in mind, but this seems to me quite realistic as an estimate but also rather modest relative to the Angst I am seeing? Not quite “ a Gilda Radner thing https://www.youtube.com/watch?v=OjYoNL4g5Vg ,” but if nothing else shows the usefulness of trying to be concrete. Here is an estimate of $100 billion extra for the UK annually https://www.governance.ai/research-paper/estimating-global-yearly-cybercrime-damage-costs . That is estimated as a twenty percent increase in UK cyber costs. Also strikes me as plausible. Brad Carson uses AI agents hurrah to produce estimates https://x.com/bradrcarson/status/2086952904682885612 . He thinks I disagree with him, but I do not. I just want to see exact numbers, and presented as a share of gdp, and then everybody can calm down just a wee bit. And here are some superforecaster estimates https://leap.forecastingresearch.org/reports/wave9 financial-losses-from-cybercrime . On this topic, I am seeing so many mistakes in financial economics across social media, but I will not go through them all. It suffices to note that very smart people have their minds turn to mush when this topic comes up, especially the “sell short” side of the matter. There is also the “beware isolated demands for rigor” response. I should not have to demand anything. If you are pretty worried about AI, you should be dying yes pun intended to provide that rigor, whether or not it is provided elsewhere. How about a new policy?: “All AI safety expressions of despair, grief, panic, anger, or insinuations thereof, with or without gifs, shall be accompanied by an estimated percentage of gdp footnote.” Would inject some real sanity into the discussion and drain out some of the all too rampant emotions.