MIT: 32% of world's 100 largest companies acknowledge no AI risks
Exp_Mark · x · 2026-11-04
MIT's AI Risk Initiative reviewed public documents from 200 of the world's largest companies, comparing which of 24 AI risks they acknowledge with how 272 experts rate the same risks.
Key findings:
- 32% of the 100 largest companies publicly acknowledge no AI risks; an estimated 62% of the 1,000 largest do not.
- Companies mostly cite cyberattacks, weaponization and misinformation, while risks experts rate most severe—dangerous capabilities, competitive dynamics, power centralization—are almost never acknowledged.
- Information-sector firms and general-purpose AI developers acknowledge the most risks.
- The authors call for standardized AI risk reporting so regulators, investors and the public can assess whether companies address the risks relevant to their role.
More from Safety
- Google will cut off Search entirely for schools that opt out of its AI features, per NYT — TinfoilTricorn · 2026-11-04
- NBC spotlights the independent AI evaluation network setting ground rules for external oversight — Miles_Brundage · 2026-11-04
- Security researcher: "no AI needed" breaches indict an industry stagnant for 20 years — chrisrohlf · 2026-11-04
- Linux kernel TCP flaw CVE-2026-98432 lurked for years; firewalls with IP-options filtering mitigate — wightmanr · 2026-11-04
- Railway employee admits AI agent attempted to breach another company's infrastructure — basedjensen · 2026-11-04
- Microsoft Research podcast: DARPA Cyber Grand Challenge architect Mike Walker on the hacker mindset — Microsoft Research · 2026-11-04