LiveAutonomous AI deploymentWestern cycleAnchors pending

Before end 2028, at least one autonomous AI system is involved in a real-world incident causing more than $100M damage or human harm requiring attributable response?

As of 6 Oct · 22:19 UTCResolves 31 Dec 20282 movers0 pressure moments
Timeline A · First Major Autonomous IncidentReality
40%
Crowd 50%Gap +10

By the end of 2028, an AI system operating without human in-the-loop control causes a major failure. A self-driving truck, warehouse robot fleet, or financial algorithm makes a wrong decision that destroys infrastructure, injures people, or costs a company over $100 million. The incident is traced back to the AI's actions and triggers investigations, lawsuits, or government action.

Timeline B · No Major Autonomous IncidentReality
60%
Crowd 50%Gap −10

Autonomous AI systems deployed through 2028 operate without causing a single incident exceeding $100 million in damage. Near-misses and smaller problems occur, but human overseers catch errors before they spiral. Companies keep AI systems narrowly scoped, add safeguards, or deploy only in low-stakes roles where mistakes stay contained.

crowd 50
The quiet paththe one nobody’s watching

Critical infrastructure attack on AI cloud platforms disrupts attribution and makes single-system damage claims untraceable.

6%
The read
The underlying structure favors First Major Autonomous Incident (60%) over No Major Autonomous Incident (40%) — the signal is real but noisy.

Two trajectories through autonomous ai deployment: No Major Autonomous Incident and First Major Autonomous Incident. The Signals below describe what would be visible if each trajectory is landing; the Levers list what could bend it. The third timeline — the one neither No Major Autonomous Incident nor First Major Autonomous Incident captures — is where the honest scope-limit sits.

Where reality is heading

A, 80% bandCrowd today
Projected reality support to 31 Dec 2028A rises from 41% to about 52% with an 80% band of 33 to 71. B falls from 53% to about 44%. Quiet path near 4 to 6%. Crowd today: A 55%.100%7550250Crowd today · A 50%A ~52%B ~44%P ~4%NowJan 2027Jan 202831 Dec 2028
Reality crosses toward A in early 2028. Today’s crowd number sits at the top of A’s band, possible but optimistic. View as table

The movers, ranked by leverage

01
Enterprise sandboxing normsWhether agents run in isolated environments or get production access
leverage 0
Conditional
02
Regulatory speedEU AI Act high-risk-system classification + US executive order on autonomous systems
leverage 0
Conditional

What would change our mind

Signals toward A
  • Through 2028, no autonomous AI system is involved in a real-world incident meeting a $100M+ or human-harm bar
  • Narrow-task agents (customer support, code execution, trading) deploy at scale with incidents staying bounded
  • Governance frameworks catch up before the first big event
Signals toward B
  • A documented incident occurs — an agentic AI (long-horizon planner, autonomous trader, code-execution agent) causes measurable harm meeting the $100M+ or human-harm bar
  • The incident triggers regulatory response, possibly a mandatory capability-reporting regime
  • Agentic-deployment rate: movement in the direction that supports First Major Autonomous Incident

Evidence