MIT Study Reveals Significant Gap in AI Risk Disclosure

1 reports archived1 independent sourcesupdated 11/5/2026, 08:58:40

Synthesis & Latest Updates

1 Sources Cross-Validated

A joint study by MIT and the University of Queensland reveals that most large tech companies fail to disclose AI risks, with severe threats like power concentration rarely mentioned. Developers prioritize reporting discrimination over alignment failures, prompting calls for transparent risk management frameworks.

LATEST/A joint study by MIT and the University of Queensland reveals that most large tech companies fail to disclose AI risks, with severe threats like power concentration rarely mentioned. Developers prioritize reporting discrimination over alignment failures, prompting calls for transparent risk management frameworks.

Hourly heat curve

2 fully observed hours

Fewer than 3 fully observed hours; trend not plotted yet.

Coverage timeline

Total 1 reports · Latest first

Rohan Paul (@rohanpaul_ai)T2Nov 5, 08:58·68 pts
MIT Study Reveals Significant Gap in AI Risk Disclosure
- 32% of top 100 companies disclosed no AI risks; estimated at 62% for the top 1000
- Severe risks like power concentration are rarely mentioned; only 4 top 100 firms cited dangerous AI capabilities
- Developers report discrimination (85%) far more frequently than misalignment (15%)
View Related Reports & Analysis Original Source