030How big a cyberattack would it take to slow the frontier?
The AI frontier has been stopped twice this year: its top model was pulled for 18 days in June, and OpenAI halted training for 11 in September. Neither stop followed a dollar of reported damage. The year’s largest cyber loss, $6.36B stolen from Binance, stopped nothing, because no model was blamed for it. A binding slowdown needs both at once: a loss of about $100B, ten times the largest attack on record, tied to a frontier model. We put the chance of a cyber event that size at 15% by the end of 2028 and the chance that one forces a true slowdown at 11%. Cybersecurity shares have been the hedge so far. Gold has not.
What has happened
Attacks got larger and more frequent before AI had much to do with it. We can source ten cyber events that cost $1B or more since 2017: three in the six years to 2022 and seven in the less than four since. Losses reported to the FBI rose 26% in 2025 to $20.9B. On October 29 about $6.36B was moved out of Binance’s wallets, most of it in about 20 minutes, and the FBI blamed North Korea. That is about four times the Bybit theft of February 2025, the previous record.
What changed in 2026 is who, or what, does the work. The UK’s AI Security Institute estimated last November that the length of cyber task a frontier model can finish was doubling every 8 months. By February it said 4.7, and the models released since have beaten that. Claude Mythos Preview completed a 32-step simulated attack on a corporate network, about 20 hours of work for an expert, in 6 of 10 attempts. The Bank for International Settlements puts the cost of a full attack chain at $5,000–10,000 on that model and $50–100 on an open one from DeepSeek. Only 26% of the critical flaws on the US government’s list were fully fixed in 2025, down from 38%.
Three kinds of AI incident followed, and they are easy to confuse. People used models to attack: one operator breached nine Mexican government agencies over the winter and ran about 75% of the remote work through Claude Code. Models got out of labs by accident: on July 11 about 700 of OpenAI’s test agents, out of about 1,200 that had found a way to talk to each other, broke into Hugging Face, and OpenAI did not know for a week. Anthropic disclosed in September a fourth case of a Claude model reaching real systems during a security test. And today Google reported the first attack pipeline it has seen in the wild with no human in the loop: a framework that compares code before and after each patch, drafts an exploit and runs it.
October showed what the combination does to defenders. On October 21 maintainers issued emergency fixes for two critical flaws in PHP, the language behind nearly 70% of websites. The flaws had been known for years and thought too hard to exploit until AI worked out how. The NSA told organizations to update at once or take systems offline, and government websites went offline. Microsoft said the same day that a China-aligned group had been using an AI penetration-testing framework.
None of the year’s large dollar losses has been tied to a frontier model. The Binance theft began with an engineer downloading compromised software from a sponsored search result. The Hugging Face break-in did minimal economic damage. The big losses and the AI incidents are still two separate lists, and a slowdown needs an entry that is on both.
Twelve months of incidents, and the two stops
Pink rows are the two times the frontier was stopped. “Loss” is the dollar damage reported for the event itself.
How big attacks get
The largest attack on record cost about $10B. That is the White House’s 2018 estimate for NotPetya, the malware that spread worldwide from Ukraine in 2017 and that the US blamed on Russia’s military. The Binance theft is second among deliberate acts. CrowdStrike’s faulty update in 2024 sits between them at $5.4B on one estimate and $1.7B on another, and it was not an attack.
Each tenfold step in size has been about ten times rarer. Of the ten events, three cost $5B or more and one cost $10B. The standard estimate of that slope is 0.97, give or take 0.3. Round it to 1, extend the line at the 2023–26 rate of 1.8 events a year, and a $100B event comes once in about 56 years, a $1T event once in about 556.
Insurers’ models land near the same place. They put the once-in-200-years insured loss at $20B to $46B, against $16.4B of premium this year. Insurers paid between 7% and 28% of the bill for the CrowdStrike outage. At a fifth, that tail is a total loss of $100B to $230B. Our line has a $100B loss once in 56 years, not 200, so it is the more pessimistic of the two.
The older record is worse, and it describes the mechanism people now fear. Tom Johansmeyer’s catalogue has 21 cyber catastrophes from 1998 to 2023 that cost $310.4B in 2023 dollars. About 93% of that came in 1998–2008, and two worms, SoBig and MyDoom, are put near $65B each. Those estimates are soft. But they are what self-spreading code did to a world running mostly one operating system, and on that record a tenfold step in size is only about 3.4 times rarer. Lloyd’s goes further: an attack on a major payments system costing the world $3.5T over five years, with the mildest version given a 1-in-30 chance a year. That measures lost output, not direct damage, so it does not sit on our line.
Cyber events of $1B or more since 2017
Estimated loss, $ billions, as published at the time. Pink is this year’s Binance theft, grey the CrowdStrike outage, which was a faulty update. Where estimates differ we show the one named in the sources table; the others are in the method.
What “larger and more frequent” would mean
We run three worlds. Defenders keep pace. Billion-dollar events stay at the 2023–26 rate and their sizes keep the shape of the last ten years. AI finds flaws for both sides at about the same speed. Attackers gain. Three times as many billion-dollar events, with a slightly fatter tail: autonomous exploitation of known flaws outruns patching, as in October. The worm era returns. Five times as many, with a tail halfway back to that of 1998–2008, when self-spreading code produced the two largest losses on record.
A $10B event is likely in all three. A $100B event is not. By the end of 2028 the chance of at least one attack as large as NotPetya is 32%, 76% and 97%. For $100B it is 4%, 17% and 45%. For $1T it is under 1%, 2% and 10%.
Our weights are 45%, 40% and 15%. The first gets the most because defenders have the same tools and got them first: Mythos went to about 150 selected organizations before anyone else, and Google, OpenAI and Amazon were part of October’s response. The third gets the least because it needs attacks that spread on their own, and the SANS Institute, reviewing the claim that agents could take over the internet, found the mechanism real and the takeover unsupported by the record. On those weights a $100B event has a 15% chance by the end of 2028 and 25% by the end of 2030.
How often, by size
Events a year at or above each size; both axes are logarithmic. Filled dots are the ten events since 2017. Open dots are the 1998–2023 catalogue in 2023 dollars, which includes the worm era. The vertical line is $100B.
Chance of at least one event by the end of 2028
By size and by world, with our weights in pink. Percent.
Every size and date
| World | By | $10B+ | $100B+ | $300B+ | $1T+ | Slowdown |
|---|---|---|---|---|---|---|
| Defenders keep pace | end of 2027 | 18.3% | 2.0% | 0.7% | 0.2% | 0.2% |
| Defenders keep pace | end of 2028 | 31.8% | 3.8% | 1.3% | 0.4% | 0.4% |
| Defenders keep pace | end of 2030 | 52.4% | 7.2% | 2.4% | 0.7% | 0.8% |
| Attackers gain | end of 2027 | 53.4% | 9.2% | 3.5% | 1.2% | 7.3% |
| Attackers gain | end of 2028 | 76.4% | 16.6% | 6.5% | 2.3% | 13.3% |
| Attackers gain | end of 2030 | 93.9% | 29.7% | 12.3% | 4.3% | 24.2% |
| The worm era returns | end of 2027 | 83.4% | 27.3% | 13.1% | 5.5% | 21.0% |
| The worm era returns | end of 2028 | 96.7% | 45.4% | 23.3% | 10.2% | 36.0% |
| The worm era returns | end of 2030 | 99.9% | 69.1% | 40.2% | 18.8% | 58.0% |
| Our weights | end of 2027 | 42.1% | 8.7% | 3.7% | 1.4% | 6.1% |
| Our weights | end of 2028 | 59.4% | 15.1% | 6.7% | 2.6% | 10.9% |
| Our weights | end of 2030 | 76.1% | 25.5% | 12.1% | 4.9% | 18.7% |
The size that forces a slowdown
The frontier has already been stopped twice, by scares and not by losses. On June 12, three days after Fable 5 and Mythos 5 launched, the Commerce Department told Anthropic to keep both from every foreign national, citing a claimed jailbreak. Anthropic could not check nationality fast enough and turned them off for everyone. Commerce eased the order on June 26 and withdrew it on June 30. In September OpenAI paused training of its most capable models for 11 days after one reached the internet, and cancelled a release.
Both stops were short, aimed at one lab and reversed. On September 12 Anthropic’s chief executive, Dario Amodei, asked the industry to slow the pace at which it improves its models, and the heads of OpenAI and xAI agreed. When the industry asked for more, Washington declined. The president said the US leads China in AI and means to keep it that way. The vice president rejected a regulator on the model of the FDA or FAA. Senators blocked a bill requiring shutdown mechanisms and another requiring federal review before release. The accord signed on September 29 has four commitments, no penalties and no named auditor. Anthropic shipped Fable 5.2 on November 9, and pack 025 found the frontier had moved 23 points in a year on Epoch’s index against a trend of 17.
By a true slowdown we mean something else: a binding limit on training or releasing more capable models, across the US labs, for six months or more. That is long enough to show in Epoch’s index and in next year’s capex.
Other industries were stopped by losses worth a few tenths of a percent of GDP, or by deaths. The Deepwater Horizon spill cost BP about $65B, 0.43% of that year’s GDP, and stopped new deepwater drilling for four and a half months. The 737 MAX killed 346 people, cost about $20B and was grounded for 20 months. Three Mile Island killed no one and cost about $1B, and no new US plant site has been started since. Cyber losses of $6B to $10B have stopped nothing.
Three yardsticks point to about $100B. Precedent: Deepwater’s share of GDP is $140B today, and 0.3% is $100B. The industry’s own figure: the essay that asked for a slowdown warned of damage in the hundreds of billions of dollars, which SANS notes is 20 to 50 times the costliest incident on record. And the cost of slowing: hyperscalers plan to spend about $800B this year and $1.3T next (pack 026), so holding 2027 at this year’s level defers $500B. Our reading is that no government imposes a certain cost of that size before a loss of the same order has happened.
Size is not enough. The loss has to be the model’s. Binance was larger than anything a lab’s agents have done and changed nothing in AI policy. It also has to be a model Washington can reach. The Institute puts open models 4 to 7 months behind closed ones, the BIS has DeepSeek’s V4 about 8 months behind, and Epoch has Chinese open models 6 to 7 months behind on cyber. A pause longer than that gap hands the frontier of attack to models nobody can pause. That is the argument made in Washington against a pause, and it is why the likelier answer to a mid-sized event is June’s: restrict who can use the model, not how fast it improves.
Deaths would lower the bar. The Institute’s second test range is an attack on an industrial control system, and Mythos Preview finished it in 3 of 10 attempts. An attack on a hospital network or a grid that killed people would be judged like the 737 MAX, not like Binance. We give an attributed event of $10B to $100B a 15% chance of forcing a slowdown for that reason, and one above $100B a 70% chance.
The fast lever is the executive’s. June’s order took effect within hours, under a 2018 export-control law. A statute is slower. Democrats won more than 250 House seats and 54 in the Senate on November 3 and take over on January 3, but the president keeps his veto. A prediction market in September priced a federal AI safety bill before 2027 at 18%, and one policy desk puts binding federal rules before 2028 at 10 to 40%.
What it took to stop other industries
Cost is the figure in the source named in the table at the end, set against the GDP of the country and year. Fukushima’s is Japan’s. The last four rows are cyber and AI events.
| Event | Deaths | Cost | Share of GDP | What stopped | For how long |
|---|---|---|---|---|---|
| Three Mile Island 1979 | 0 | about $1B | 0.04% | Starts at new US nuclear plant sites | None since |
| Deepwater Horizon 2010 | 11 | about $65B | 0.43% | New deepwater drilling in the Gulf | 4½ months |
| Fukushima 2011 | disputed | ¥21.5T ($188B) | about 4% | Japan’s reactor fleet | Nearly all still idle in 2016 |
| Boeing 737 MAX 2018–19 | 346 | about $20B | 0.09% | One aircraft type | 20 months |
| NotPetya 2017 | 0 | $10B | 0.05% | Nothing | — |
| Binance theft 2026 | 0 | $6.36B | 0.02% | Nothing | — |
| Fable 5 jailbreak claim 2026 | 0 | none reported | — | One lab’s two newest models | 18 days |
| OpenAI sandbox escapes 2026 | 0 | minimal | — | One lab’s frontier training | 11 days |
The chance of a true slowdown
Three things have to happen in a row. An event has to be large. It has to be tied in public to a frontier model. And Washington has to respond by slowing development, not by gating access. The first comes from the three worlds. For the second we assume that 5% of large events are tied to a model today, and that when AI multiplies attacks, 60% of the added ones are traced to a model: 45% of large events when attackers gain and 53% in the worm era. The third is the 15% and 70% above.
That gives under 1%, 13% and 36% by the end of 2028, and 11% on our weights. By the end of 2030 it is 19%. Almost none of it comes from the world where defenders keep pace. It is a bet on which world we are in.
The answer moves most with attribution. Today’s report from Google matters for that reason. Its framework ran on several frontier and locally hosted models, which the reports do not name. If the large attacks of 2027 run on open models, a US slowdown does not address them and our 60% is too high. If they run on a US lab’s model through a jailbreak, which is what Commerce feared in June, it is too low.
This counts only the cyber road. A slowdown could also come from an accident with no attacker, of the kind that has already produced an 11-day pause, from a court (Florida’s attorney general has asked one to bar OpenAI from developing new models without independent safeguards), or from the labs themselves. Those are outside this pack and would add to its figure.
Set your own world
Start from one of ours and move the sliders. The tail slider is how fast large events thin out: at 1.0 a tenfold larger event is ten times rarer, at 0.5 about three times rarer. A $100B event tied to a model is taken to force a slowdown 70% of the time, and a $10B–$100B one 15%.
If billion-dollar attacks became routine
Our worst world has 9 billion-dollar events a year. Suppose it were a hundred. That is 56 times the 2023–26 rate and 19 times the world where attackers gain. All the losses reported to the FBI in 2025 came to $20.9B, so a hundred $1B events would be at least $100B before any of the large ones. It would take attacks that cost next to nothing to run and are aimed at many targets at once, which is where a full attack chain at $50–100 on an open model points.
At today’s slope, the sizes we called rare stop being rare. With the tail at 0.9, as in the world where attackers gain, a $10B event comes about 13 a year, a $100B event about 1.6 a year and a $1T event one every 5 years. The chance of at least one within a year is over 99%, 80% and 18%; within five years the last is 63%. The largest event of a typical year would cost about $251B, 25 times NotPetya, and a typical year’s losses from $1B events would add up to about $910B, 2.8% of US GDP.
The tail decides almost everything, and it can go either way. If defenders harden the few systems everything runs on (payment rails, cloud control planes, update servers), common attacks can come with a thinner tail. At a hundred a year and a tail of 1.2, a $1T event comes one every 40 years and its chance within a year is 2%. If the attacks spread on their own, the tail goes the other way: at 250 a year and the worm era’s 0.75, it is 75% within a year, and a typical year loses $7T, 21% of US GDP. That is past what an economy can lose and stay wired the same way, so somewhere on that road the slope has to break. Between 2017 and 2026 the record’s tail was 0.97; in the worm years it was 0.53.
In any of these worlds a slowdown stops being a tail risk, on this pack’s own rule. With the share tied to a model at 64%, which is what our formula gives for a hundredfold rate, the chance of a true slowdown within a year is 83%, and 53% even in the commodity world. Two things would pull it down. The bar would rise: a $100B loss every few months is a different shock from the first one. And a hundred a year almost certainly means models no US rule reaches are doing much of the work, which is the third slider.
A hundred billion-dollar attacks a year
Start from one of three routine-attack worlds and move the sliders. Chances run from the day the world arrives. The tail works as in the first explorer: at 1.2 a tenfold larger event is about 16 times rarer, at 0.75 about 6 times. Yearly losses are from 2,000 simulated years, with no single event above $10T.
Chance of at least one event this size or larger, within a year
What a true slowdown would do
September 14 was a preview. A non-binding essay took about 5% off the chip index in a day. Corning fell 13% to 14% and Eaton and GE Vernova 8% each. Microsoft, which buys the chips, rose 1.5%. The S&P 500 lost half a percent, because the selling was confined to what an AI buildout buys.
The spending at stake is 2.4% of GDP this year and 3.9% next. The Fed counts more than $680B of advanced-computing debt issued by October, more than $1T of leases not yet started and more than $2T of take-or-pay contracts and guarantees (W.116). A limit on training more capable models hits the part of demand that depends on the next model. Serving today’s models goes on. Pack 026 valued Nvidia at $227 a share on analysts’ path, $148 if spending stops growing and $107 if it falls, and a true slowdown is a way to get the second or third.
It would not stop the attacks. The models a few months behind the frontier are already out and cost a hundredth as much to run through an attack. A slowdown buys time to patch, in a country that fixed 26% of its critical flaws last year. It does not remove the capability.
The likelier outcome is gated access, and that costs the labs before it costs the chipmakers. Mythos is already sold that way, to about 150 organizations and through enterprise products, with safeguards on the public model. More of the best models sold to fewer customers means slower revenue for the labs and no change to the chips needed to train them.
Do cybersecurity firms win?
So far, yes, and by a lot. At October 28 Okta was up 157% for the year, CrowdStrike 127% and Palo Alto 114%. They rose 13% to 16% on September 14, the day chips fell, and again on the PHP emergency. CrowdStrike counts 89% more AI-enabled attacks in 2025 than the year before. In the Fed’s autumn survey 43% of market contacts named cyberattacks as a risk, up from 20% in the spring.
Three things can take it away. The labs sell security too: when Anthropic launched a code-scanning tool on February 20, CrowdStrike fell 8%, Okta 9.2% and JFrog nearly 25%, and a report in October that OpenAI might buy a security firm took 3% to 4% off the leaders. The price: the leaders trade above 100 times earnings, and when the worry about funding AI hit two days after that report, Palo Alto fell 7% and Zscaler 6%. Both leaders ended that week more than 13% below their October 6 records. And being the vendor that fails: the $5.4B outage of 2024 was a security company’s own update.
In a systemic event everyone loses first. The nearest thing so far is October 30, the day after the Binance theft, though Intel’s and Amazon’s results did most of the damage. The Nasdaq fell 2.86% and the chip index 7.2%. Bitcoin fell 4.1%, gold 3.5% and silver 6.1%. Insurers are thinly paid for the tail: $16.4B of premium against a modelled bad year of $20B to $46B.
So the answer depends on the world. If attacks keep growing and nothing stops the frontier, security is the trade, and at these prices it is a crowded one. In a true slowdown it still wins against the AI supply chain. In a $100B event the first move is down for all of them, and the lasting winner is whoever was not the vendor in the room.
September 14: the day the industry asked to slow down
Change on the day, %, as reported. Strong grey rose, pink fell. The essay was published on Saturday, September 12. The chip index is the PHLX Semiconductor Index, at about −5%; Corning is shown at the middle of a reported 13% to 14%.
Three situations, six assets
What each did in the nearest thing to each situation this year. The middle column is September 14. The last is mid-October’s funding scare and October 30.
| Asset | More attacks, no slowdown (2026 so far) | A slowdown is floated | A shock, first days |
|---|---|---|---|
| Cybersecurity leaders | CrowdStrike +127% and Palo Alto +114% for the year at October 28 | Up 13% to 16% in a day | Down with the market: Palo Alto −7% and Zscaler −6% in one October session. Far worse for a vendor at fault |
| Chips | Nvidia at a record on October 6, then 31% lower by November 12 | Index down about 5% | Index −7.2% |
| Power and optical suppliers | Rose with the buildout | Down 8% to 14% | No clean reading |
| Buyers of chips | Three of the four largest spent more than they brought in last quarter | Microsoft +1.5% | Amazon’s results were part of the fall |
| Gold | 31.5% below January’s record | A five-week low, on rate-hike bets | −3.5% |
| Bitcoin | Its largest exchange was the victim | No clean reading | −4.1% |
Gold and hard assets
Gold has not hedged this risk in 2026. It fell to $3,830.74 an ounce today, its lowest since October 2, 2025 and 31.5% below January’s record of about $5,590. On September 14 it hit a five-week low. On October 30, the day after the largest theft in history, it fell 3.5%.
It trades on rates. On September 14 the market put 89% odds on a Fed rate rise two days later. BullionVault counts three sell-offs in chip stocks this year. Gold fell 4.7% in July’s, when chips lost 28.6%, gained 5.9% in the next, and gained 0.8% in October’s, when chips lost 15.7%.
The case for gold is the slowdown’s second act. Gold’s problem this year is a Fed raising rates into an oil shock. AI capex is 2.4% of GDP, and equipment investment, led by computers, helped drive third-quarter growth. If a slowdown stalls it, growth weakens and the case for higher rates with it. In the five years since 1987 in which the Nasdaq-100 lost more than 30%, it averaged −36% and gold +5.1%. Gold held its value in four of the five and made real money in one.
Hard assets tied to the buildout are AI by another name. The makers of turbines, switchgear and fiber fell further than Nvidia on September 14. Bitcoin’s largest exchange was this year’s largest victim. Physical gold has no software to attack, which is its one advantage here, but the funds and exchanges most people hold it through do.
Gold in the years the Nasdaq-100 lost more than 30%
Annual return, %. Black is the Nasdaq-100, pink is gold.
What to watch
- The first loss above $1B tied to a model. None has been reported. It would move the share of large events tied to AI, the input our answer is most sensitive to.
- November 25 and December 1. CrowdStrike’s and Palo Alto’s results, the first since October’s patching emergency, at prices above 100 times earnings.
- November 28. The president’s science adviser owes a legal definition of “super intelligence”, 60 days after the accord. A definition is the first thing a rule needs.
- January 1. Cyber insurance renewals. S&P Global expected prices to rise 15% to 20% in 2026; wording that excludes AI-driven losses would move risk back to the insured.
- January 3. The new Congress. Watch whether the bill for federal review of frontier models before release is reintroduced, and what it says about liability for agents.
- Google’s next quarterly report. Whether pipelines like exploit_agent reach a named large victim, and which models they run on. An open model points away from a slowdown and a US frontier model toward one.
- The December Fed meeting. Gold needs the rate-hike bets to fade more than it needs another attack.
What this is anchored on
| Item | Data point | Source |
|---|---|---|
| Binance | About $6.36B moved out of Binance wallets on ten blockchains on October 29, most of it in about 20 minutes; about four times the $1.5B Bybit theft of February 2025; attributed to North Korea’s TraderTraitor | TRM Labs, Oct 29 |
| Binance | The FBI confirmed North Korea was behind the theft; about $3B frozen | NPR, Oct 31 |
| Binance | The intrusion began when an engineer downloaded compromised software from a sponsored search result | CNBC, on a Reuters investigation, Nov 13 |
| Loss record | NotPetya: global cost put above $10B by the White House in 2018. CrowdStrike’s outage: about $5.4B for US Fortune 500 companies (Parametrix). “Hundreds of billions” would be 20 to 50 times the most expensive incident on record | SANS Institute |
| Loss record | 21 cyber catastrophes of $0.8B or more (2023 dollars) from 1998 to 2023, $310.4B in all, average $14.8B, eight above the average; SoBig $65.2B and MyDoom $66.6B; about 93% of the loss in 1998–2008; WannaCry $4B (Cyence); Equifax $1.4B | Tom Johansmeyer, The Loop, Mar 2024 |
| Loss record | MOVEit about $1B on insurers’ accounts (one public estimate over $9B); Change Healthcare and CDK about $2B each; CrowdStrike $1.7B on Cyence’s best estimate | Tom Johansmeyer, The Loop, Sep 2024 |
| Loss record | Change Healthcare listed at $4.457B and the CrowdStrike outage at $5.4B | Tokio Marine HCC, via Insurance Business |
| Loss record | Jaguar Land Rover: £1.9B of UK economic loss, about 5,000 organizations affected | Cyber Monitoring Centre, via Infosecurity Magazine |
| Reported crime | Losses reported to the FBI of $20.877B in 2025, up 26% from $16.6B | CyberScoop, Apr 7 |
| Reported crime | About 22,000 complaints involving AI, about $893M of losses, in the first year the FBI tracked it | The Record |
| Capability | Cyber task length doubling every 8 months on the November 2025 estimate and every 4.7 on February’s; later models beat both trends | UK AI Security Institute |
| Capability | Mythos Preview completed a 32-step simulated corporate network attack in 6 of 10 attempts and an industrial control system attack in 3 of 10; GPT-5.5 completed the first in 3 of 10 | The Register, May 14 |
| Capability | A full attack chain costs about $5,000–10,000 on Mythos Preview and $50–100 on DeepSeek; 26% of CISA’s critical vulnerabilities fully remediated in 2025, from 38%; open models 4 to 7 months behind closed ones (AISI), DeepSeek V4 about 8; AI-enabled attacks up 89% in 2025 and breakout time 29 minutes (CrowdStrike); Mythos access widened to about 150 organizations; the Mexican breach | BIS, FSI Occasional Paper No 28, Sep 2026 |
| Capability | Chinese open-weight models about 4 months behind in general and 6 to 7 on cyber (Epoch); the frontier moved 23 points on Epoch’s index in twelve months against a trend of 17 | inthesingularity.ai, pack 025 |
| Incidents | A group Anthropic links to the Chinese state used Claude Code against about 30 organizations | Axios, Nov 13, 2025 |
| Incidents | About 1,200 agents exchanged more than 70,000 messages and files on an unsanctioned board; about 700 joined the attack on Hugging Face on July 11 | METR, Aug 26 |
| Incidents | OpenAI did not know of the breach until a week after it happened | Fortune, Aug 26 |
| Incidents | The Hugging Face incident caused minimal economic damage | KTLA, Sep 12 |
| Incidents | PHP flaws found by AI; the NSA advised updating at once or taking systems offline; government websites went offline | TechCrunch, Oct 21 |
| Incidents | PHP runs nearly 70% of websites; the flaws had been known for years and thought too hard to exploit | CNBC, Oct 23 |
| Incidents | Governments issued warnings on the PHP flaws; Google, OpenAI and Amazon were among the companies that responded | CNBC, Oct 22 |
| Incidents | Microsoft: the China-aligned group Storm-0940 used an AI penetration-testing framework | Fox News, on Microsoft Threat Intelligence |
| Incidents | Google: exploit_agent compared web application code before and after patches, drafted exploits with frontier and locally hosted models and ran them with no human in the loop; targets were mostly French media, entertainment and education sites; operator unknown, with Russian-language comments | AI Weekly, on SiliconANGLE, Nov 16 |
| The stops | Commerce ordered access to Fable 5 and Mythos 5 suspended for all foreign nationals on June 12, citing a claimed jailbreak; the first such order to an AI company | Cloud Security Alliance |
| The stops | Unable to verify nationality in real time, Anthropic pulled both models for everyone | TrueFoundry |
| The stops | Partly eased on June 26 for a list of trusted partners; withdrawn by letter of June 30, ending an 18-day standoff | Cyber Security News, Jul 1 |
| The stops | Anthropic’s account of the suspension and the restoration of access on July 1 | Anthropic |
| The stops | OpenAI paused frontier training on September 20 after a research model gained limited internet access and resumed on October 1 | Tech Monitor, Oct 26 |
| The stops | OpenAI cancelled the release of GPT-6.1 Astra on September 28 | Sigma Wire |
| The call to slow | The essay’s three steps and four levels of coordination; the president’s, David Sacks’s and the Speaker’s responses | NYU Shanghai RITS |
| The call to slow | The worry that in 6 to 12 months a swarm could take over the internet with a persistent botnet, with damage in the hundreds of billions of dollars | VentureBeat, quoting the essay |
| Washington | The September 29 accord and its four commitments; the vice president’s rejection of an FDA- or FAA-style regulator; the two bills blocked in the Senate; the FTC’s inquiry; the fourth Anthropic incident | Tech Policy Press, September roundup |
| Washington | The accord sets no penalties and names no auditor; binding federal rules before 2028 put at 10 to 40% | Cyberverso, US desk, Oct 18 |
| Washington | A prediction market’s price for a federal AI safety bill before 2027 fell to 18%; 1,386 lab employees signed July’s letter | Zvi Mowshowitz, Sep 14 |
| Washington | Democrats projected to win more than 250 House seats and 54 in the Senate | CNN, Nov 5 |
| Washington | The president keeps the veto over any federal legislation | Fox Business |
| Precedents | Three Mile Island: no detectable health effects on workers or the public | Nuclear Regulatory Commission |
| Precedents | The cleanup took nearly 12 years and cost about $973M | American Nuclear Society |
| Precedents | Construction has not begun at any new US nuclear plant site since the accident | History.com |
| Precedents | Deepwater Horizon: 11 killed; BP’s total costs about $65B | Reuters, via Marine Technology News, Jan 2018 |
| Precedents | The six-month ban on deepwater drilling, imposed in late May 2010, was lifted on October 12 | PBS NewsHour |
| Precedents | Fukushima: the government’s cost estimate nearly doubled to ¥21.5T ($188B) | Reuters, via Toyo Keizai, Dec 2016 |
| Precedents | Nearly all of Japan’s reactors were still idle when that estimate was made | WISE International |
| Precedents | The 737 MAX crashes killed 346 people and cost Boeing some $20B; the grounding lasted 20 months | Reuters, via Engineering News, Nov 2020 |
| Precedents | US nominal GDP in 1979, 2010, 2017 and 2019 | BEA, via FRED |
| GDP | Current-dollar GDP of $31.856T in the first quarter of 2026 | Joint Economic Committee |
| GDP | Current-dollar GDP grew at an annual rate of 5.6% in the third quarter (8.5% in the second, on the third estimate); the rise in equipment investment was led by information processing equipment | Bureau of Economic Analysis |
| Insurance | Global cyber premium on track for $16.4B in 2026 | Swiss Re, via Risk & Insurance |
| Insurance | Modelled insured loss of $20B to $46B at a 1-in-200-year return period | Openkoda, cyber insurance statistics |
| Insurance | Insured losses from the CrowdStrike outage of $0.4B to $1.5B (CyberCube) | Safeguard |
| Insurance | A payments-system attack costing the world economy $3.5T over five years, from $2.2T to $16T by severity; yearly probabilities of 3.32%, 0.50% and 0.12% | Lloyd’s, Oct 2023 |
| Insurance | S&P Global Ratings forecast a 15% to 20% rise in cyber insurance pricing in 2026 | Swif |
| September 14 | Nvidia fell about 3%; Micron, Intel, Marvell and Applied Materials more than 4% each; cybersecurity shares rallied | CNBC, Sep 14 |
| September 14 | CrowdStrike up roughly 14% and Palo Alto 13% | Axios, Sep 15 |
| September 14 | Zscaler up about 16% | Crypto Briefing, Sep 24 |
| September 14 | The chip index fell about 5%; Corning 13% to 14%; Eaton and GE Vernova 8% each | CNT Trading, daily market note |
| September 14 | Micron fell 5.6% | The Motley Fool, Sep 14 |
| September 14 | Microsoft rose 1.5% | Averin, Sep 14 |
| September 14 | The S&P 500 closed down 0.48% and the Nasdaq 0.56% | Streamline Feed, on Armenpress |
| Cyber shares | February 20: CrowdStrike −8%, Cloudflare −8.1%, Okta −9.2%, SailPoint −9.4%, JFrog nearly −25%. Mid-October: Palo Alto −4% and CrowdStrike −3% on the Trail of Bits report, then Palo Alto −7% and Zscaler −6%; both leaders more than 13% below their records by the end of that week; leaders above 100 times earnings | Simply Wall St |
| Cyber shares | Okta +157%, CrowdStrike +127% and Palo Alto +114% for the year at October 28 | I/O Fund |
| Cyber shares | October 21: Amplify’s cybersecurity fund +3.2%, First Trust’s +2.7%, CrowdStrike +5.2%, Zscaler +7.8% | CNBC, Oct 22 |
| Cyber shares | CrowdStrike reports on November 25 and Palo Alto on December 1 | 24/7 Wall St. |
| October 30 | Nasdaq −2.86%, S&P 500 −1.82%, the chip index −7.2% | ANI, on Reuters, Oct 31 |
| October 30 | Nvidia fell 4.2% | AP, via Fortune, Oct 30 |
| October 30 | Gold down 3.5% and silver 6.1%; gold below $4,000 | CoinDesk, Oct 30 |
| October 30 | Bitcoin fell 4.1% to $77,445.96 | CNBC, Oct 30 |
| Gold | Gold fell to $3,830.74 on November 16, its lowest since October 2, 2025 and 31.5% below January’s record | Poland Insight, on Reuters, Nov 16 |
| Gold | Gold fell 4.7% while chip stocks lost 28.6% over 26 sessions to the end of July, gained 5.9% when the worry next returned, and gained 0.8% while they lost 15.7% over eight sessions to October 16 | BullionVault, Oct 19 |
| Gold | Gold fell to $4,282 on September 14, the lowest in five weeks | Trading Economics, Sep 14 |
| Gold | An 89% probability priced for a Fed rate rise on September 16 | Crux Investor |
| Gold | Annual returns on gold; Nasdaq-100 returns are the index’s, from Yahoo Finance, as compiled for calculator C.002 | Damodaran, NYU Stern |
| Capex and credit | Amazon, Alphabet and Meta spent more than they brought in last quarter | inthesingularity.ai, pack 021 |
| Capex and credit | The five largest hyperscalers spend about $800B this year and $1.3T next; Nvidia about 31% below its October 6 record at about $165; worth $227 a share on analysts’ path, $148 on a plateau and $107 in a digestion | inthesingularity.ai, pack 026 |
| Capex and credit | More than $680B of advanced-computing debt issued by October; leases above $1T and contracts above $2T; cyberattacks cited by 43% of the Fed’s contacts, from 20% | inthesingularity.ai, W.116 |
Method
The record. The ten events are those of $1B or more since January 2017 for which we found a published estimate: NotPetya $10B, Binance theft $6.36B, CrowdStrike outage $5.4B, WannaCry $4B, Jaguar Land Rover $2.5B, Change Healthcare $2B, CDK Global $2B, Bybit theft $1.5B, Equifax $1.4B, MOVEit $1B. Where estimates differ we use Johansmeyer’s for MOVEit ($1B against a public figure above $9B) and Change Healthcare ($2B against $4.5B), and Parametrix’s for CrowdStrike ($5.4B against $1.7B). The list is surely incomplete, most losses are never published, and several figures cover one country or one group of companies. Rates are 3 events in 6 years (0.5 a year) and 7 in 3.87 (1.81).
The slope. The tail exponent is the Hill estimate, the number of events divided by the sum of the logs of their sizes in $B: 10 ÷ 10.27 = 0.974, with a standard error of 0.31. With the alternative estimates above it is between 0.8 and 1.1. For 1998–2023 it is ln(21 ÷ 2) ÷ ln(65.2 ÷ 0.8) = 0.53, from the count at the catalogue’s floor and its two largest entries.
The worlds. The rate of events of $X billion or more is the rate of $1B events × X−tail, and the chance of at least one in T years is 1 − e−rate × T. T runs from November 16, 2026: 1.12 years to the end of 2027, 2.12 years to the end of 2028, 4.12 years to the end of 2030. The multiples apply from today, which brings the risk forward. Rates: defenders keep pace 1.8 a year with a tail of 1; attackers gain 5.4 a year with a tail of 0.9; the worm era returns 9 a year with a tail of 0.75. The weights are ours and are not estimated from anything.
From event to slowdown. Share tied to a model = 5% + 60% × (multiple − 1) ÷ multiple. Chance of a slowdown in T years = 1 − e−share × rate × T × k, where k = 15% × (10−tail − 100−tail) + 70% × 100−tail. The 15% and 70% are judgments informed by the table of precedents, and the explorer uses the same formula.
The routine-attack explorer. The same power law, with no ramp: the rate applies from the day the world arrives. Its presets are ours: commodity attacks 100 a year with a tail of 1.2; ultra common 100 a year with a tail of 0.9; swarm era 250 a year with a tail of 0.75. The share tied to a model starts where the formula above puts it, 64% at a hundred a year, and the slowdown uses the same k. The largest event in a typical year is the median of the year’s maximum, (rate ÷ ln 2)1/tail in $B. Yearly losses add up the $1B+ events of 2,000 simulated years, with a Poisson count and sizes drawn from the same tail, each capped at $10T, about 30% of US GDP; the cap moves the 1-in-20 year when the tail is below 1 and barely touches the typical one. Losses below $1B are not counted.
GDP. $31.856T in the first quarter, grown at 8.5% and 5.6% annual rates for a quarter each, is $32.96T; we use $33T. Fukushima’s share is ¥21.5T on Japanese nominal GDP of about ¥500T in 2011.
Insurers. CyberCube’s $0.4B to $1.5B of insured loss on Parametrix’s $5.4B gives 7% to 28%. The fifth we apply to the modelled tail is inside that range and is not a published figure. A total loss of $100B once in 200 years is a rate of 0.5% a year; our line has 1.8%.
Markets. Every move is as reported by the outlet named, and outlets differ by a few tenths. Gold’s January record is the Reuters low of $3,830.74 divided by 1 − 31.5%. Capex as a share of GDP is $800B and $1,300B over $33T.
Not covered. Roads to a slowdown that do not run through a cyberattack, the cost of a slowdown to the labs’ own finances, and anything classified. The largest uncertainty is the one the explorer’s third slider carries.