Year of the First Major AI Harm Event
Year of the First Major AI Harm Event
If such an event happens at all, experts expect it by 2035.
Hover to highlight a group, click to show or hide itTap a group to show or hide it
Our Analysis
How do we interpret these results
Conditional on such an event occurring by the end of 2050, experts and superforecasters expect it to arrive in the mid-2030s. The median expert and superforecaster each assign a 50% chance that the first qualifying event occurs by 2035, while the public is somewhat later at 2037. Expert-public differences are statistically significant across the elicited distribution, from the 5th through 95th percentiles. Experts and superforecasters, by contrast, give closely similar forecasts and do not differ significantly at any elicited percentile. Uncertainty remains substantial: the median expert gives a 25% chance that the first event occurs by 2031 and a 75% chance by 2040, with a 5% chance by 2028 and a 95% chance by 2048.
Forecasts by group
The year by which a typical forecaster gives the event each chance of having happened
| group | 5% | 25% | 50% | 75% | 95% |
|---|---|---|---|---|---|
| Experts | 2027.6 | 2031 | 2035 | 2040 | 2048.4 |
| Superforecasters | 2028 | 2031 | 2035 | 2041 | 2049 |
| Public | 2028 | 2032 | 2037 | 2042 | 2049 |
Click any cell to compareTap any cell to compare
* Denotes non-overlapping bootstrapped confidence intervals between group aggregates.
Rationale analysis
The main lines of reasoning driving the spread in estimates
Early-resolution
respondents often argue that this could resolve soon given that AI is already sufficiently capable and integrated into critical systems: “The military capability to do this already exists, it’s just a question of when a state decides it is acceptable to use completely AI powered targeting.”; “This question will almost certainly resolve by the end of 2027 as a result of development of autonomous drones in their war with Russia.” Others reason incidents of this nature are likely to occur when the technology is still nascent, i.e. “while there is a considerable gap in our understanding of the system” and “before protective practices have a chance to develop.” As one writes, “An analogy would be a high rate of airplane incidents and accidents as a percentage of total, in the early years of rapid growth of the industry.”
Late-resolution
respondents typically point to the need for greater autonomy and deployment to materialize before an incident becomes likely: “Widespread full autonomy and agentic deployment will ramp up gradually through the 2030s rather than arriving suddenly.”; “Deployment into safety-critical systems is still relatively limited.” Several expect humans to remain in the loop where stakes are high—“AI systems will continue to operate with substantial human oversight, especially in domains where failures could cause large-scale harm”—with one adding that this dynamic means near-term incidents will likely be “human-using-AI rather than AI-as-primary-agent.”
Rationale examples
High-quality rationales from forecasters across the range
Pessimists
A single airliner carries far more than 50 people, so once flight is genuinely AI-controlled, one crash blamed on the AI clears it, and the same logic applies to autonomous vehicles, drones, or military systems. Widespread autonomous AI produces exactly these situations, so I see a qualifying event as all but certain by 2050.
I do not think the most likely pathway is ‘rogue superintelligence.’ I think the much more plausible mechanism is ordinary human overconfidence combined with increasing delegation of authority to systems that appear highly competent inside narrow operational contexts. Humans are extremely susceptible to capability theater. If a system consistently outperforms experts in some measurable domain, organizations will rapidly increase the amount of trust, autonomy, capital access, or operational responsibility granted to it. History suggests that institutional incentives almost always outrun institutional caution.
We don't have the right incentives and safeguards in place for companies to build safe products. Unclear liability rules, no third-party testing, intense competition causing a race to the bottom.
Conditional on such an event occurring by 2050, I expect it to occur relatively early in the window because (a) AI deployment in safety-critical applications is accelerating now, (b) the financial damage threshold of $100B is reachable via large-scale cyber or market events that are already plausible with current and near-term capabilities, and (c) historically novel technologies tend to cause their first major incidents within the first decade or two of widespread deployment.
Optimists
I think people are overstating the likelihood that AI acts independently of humans. [To prevent] an event this large, I would expect there to be multiple layers of human oversight, validation, and safety controls in place. AI could absolutely contribute to accidents, cyber incidents, or other harmful events, but I think humans will still be heavily involved in any system capable of causing this much damage. Because the question requires the event to be primarily driven by AI, I think the probability remains very low, though not zero.
The resolution criteria matter enormously here, it seems quite likely that human negligence causes 50 deaths (this is fairly common) while enabled/augmented by AI, but for AI to be the primary cause [is unlikely].