Incident Tracker - June 2025 Update
A freely available, continuously updated AI incident dataset with structured NatSec and harm categorisation gives APS risk and governance practitioners a concrete evidence base for AI risk work.
Key points
- MIT AI Risk Repository's incident tracker has been updated to include all AIID incidents through 23 June 2025 (up to ID #1116).
- New features include national security impact assessment across five categories, harm severity rescaling, and Fishbone causal diagrams.
- Useful as a reference dataset for APS agencies developing AI risk registers or incident classification frameworks.
Implications for Australian agencies
- Consider Agencies developing AI risk registers or incident classification frameworks could consider referencing the MIT tracker's taxonomy and harm severity scale as an external benchmark.
- Monitor AI governance and national security policy teams may want to monitor the NatSec impact assessment feature as a potential input to whole-of-government AI risk horizon scanning.
Implications are AI-generated. Starting points, not advice — see methodology for how they're framed.
View original source
Copied.
"Incident Tracker - June 2025 Update"
Source: MIT AI Risk Repository – Blog
Published: 18 July 2025
URL: https://airisk.mit.edu/blog/incident-tracker---june-2025-update
The MIT AI Risk Repository has released a June 2025 update to its AI incident tracker, synchronising with the AI Incident Database through incident ID #1116. The update introduces a simplified 1–5 harm severity scale, spider-chart impact profiles, and a new national security impact assessment covering physical security, information warfare, sovereignty, economic security, and societal stability. Additional analytical features include Fishbone causal diagrams, ambiguity flagging, and classification of the primary goal of AI systems involved in each incident. The full dataset has been reclassified under the updated methodology for consistency.
Implications for Australian agencies:
- [Consider] Agencies developing AI risk registers or incident classification frameworks could consider referencing the MIT tracker's taxonomy and harm severity scale as an external benchmark.
- [Monitor] AI governance and national security policy teams may want to monitor the NatSec impact assessment feature as a potential input to whole-of-government AI risk horizon scanning.
Retrieved from SIMS, 18 July 2026.