AI Risk Wire

AI Risk Monitor

How attacks get in: the first step · Last 12 months

How we count

The first step of the attack path of each real-world incident made public in Nov 2025 – Oct 2026, where the sources state it. An incident counts once.

Breaking in 130Running malicious commands or prompts 111Getting access to the AI model 76Causing damage or disruption 52Gathering data 45Preparing tools and accounts 39Preparing the attack on the AI 38Scouting the target 37Avoiding detection 25Stealing passwords and keys 10Sending data out 9Exploring systems once inside 6Keeping a foothold 3Gaining more permissions 3Controlling it remotely 2Spreading to other systems 1

587 of 632 incidents state it.

Running malicious commands or prompts: 22% → 13%

% of real-world reports: ○ Feb–Jul ● Aug–Oct · largest change, not a trend

0%10%20% Breaking in19% → 20%Running malicious commands or prompts22% → 13%Getting access to the AI model13% → 12%Gathering data7% → 8%Causing damage or disruption9% → 8%Preparing tools and accounts6% → 7%Preparing the attack on the AI5% → 6%

How the first attack step is changing

stated in 296 of 318 incidents, last 3 full months

First attack stepLast 3 full months6 months before
Breaking in78 of 29653 of 273
Running malicious commands or prompts41 of 29663 of 273
Getting access to the AI model33 of 29636 of 273
Causing damage or disruption18 of 29624 of 273
Gathering data23 of 29619 of 273
Preparing tools and accounts20 of 29620 of 273
Scouting the target20 of 29615 of 273
Preparing the attack on the AI19 of 29614 of 273

What caused the harm: AI used to attack, AI attacked, user misuse, AI acting alone · Last 12 months

How we count

How the harm came about, as classified for each real-world incident made public in Nov 2025 – Oct 2026. Incidents not yet classified are not counted.

AI used to exploit 210AI attacked 108Misuse by its own users 89AI failing in normal use 44AI acting on its own 32Harm by design 27Supply chain attack 13

523 of 632 incidents state it.

Misuse by its own users: 18% → 9%

% of real-world reports: ○ Feb–Jul ● Aug–Oct · falling

0%10%20%30% AI used to exploit29% → 35%AI attacked17% → 20%Misuse by its own users18% → 9%AI acting on its own2% → 6%AI failing in normal use7% → 6%Harm by design4% → 6%Supply chain attack3% → 1%

Built 16 Nov 2026. Periods end with the last full week or month; the week or month in progress is not counted. Bold: the five largest rows in each chart, ties included; controls by incidents where the control broke. Linked numbers open the incidents or model versions they count.

Reported incidents, not rates · 75 sources polled hourly since Sep 2026; earlier incidents come from a one-off backfill · English only · stories told twice are not yet merged · our own severity grades are not counted · built 2026-11-16T15:22:29Z

Linked numbers open the incidents they count. Updated 2026-11-16T15:22:29Z