Rumman Chowdhury
CEO of Humane Intelligence; former U.S. Science Envoy for AI; former director of Twitter's META team
Humane Intelligence
Policy and government
Chowdhury ran the White House-backed public red-teaming of AI models and served as America's first science envoy for AI. She wants independent testers to be able to probe AI systems the way safety inspectors probe cars, and thinks the real work is fixing harms already happening, not debating machine consciousness.
Declines to give a number
No number; deliberately avoids existential-risk language. "Maybe they're less exciting sounding than, can AI come alive and, will AI have feelings, and will AI set off nuclear weapons? But I think, practically and pragmatically speaking..." — prioritising documented harms (FLI Paris breakfast, December 19, 2024).Source
On timing
Not a timeline forecaster; frames the issue as evaluating deployed systems now: "What interests me is how can we create tests and evaluation to mimic a real-world situation as much as possible?" (December 2024)
SourceWhat they call for
- Third party auditing and public red teaming
- Secure model access for independent auditors
- Increased NIST funding and standards
- A 'right to repair' for AI systems
- Minimum adoption standards for high risk AI
In their own words
"We already have significant evidence of AI systems in use today that infringe upon basic civil and human rights. These issues MUST be addressed immediately." (House Oversight testimony)
13 December 2023
Advocates 'algorithmic bias bounty programs' and red teaming to test systems in real-world conditions.
19 December 2024