16 hrs ago
Fired OpenAI Researchers Urge Visible AI Reasoning and Independent Safety Reviews
Three researchers who used to work at OpenAI were recently fired.
They want the company to keep checking how its AI systems come up with answers.
They also want outside safety groups to examine those systems.
They say this matters because AI is becoming more advanced.
Their requests are aimed at helping people understand and assess possible safety risks.
Three former OpenAI researchers who were recently fired called on the company to keep studying how its AI models reach answers.
They urged OpenAI to allow independent safety groups to examine its AI systems.
The researchers warned of safety risks as AI technology becomes more advanced.
Their requests concern transparency into model reasoning and outside scrutiny of AI systems.
The account does not identify the researchers or provide details about why they were fired.
- Who
- Three recently fired former OpenAI researchers.
- What
- They urged OpenAI to keep studying how its AI models reach answers and to allow independent safety groups to examine its systems.
- Where
- OpenAI; no location is specified.
- When
- Recently; no specific date is given.
- Why
- They warned of safety risks as AI technology becomes more advanced.
Researchers’ calls
Company position
Visibility into AI reasoning
Researchers’ calls
The former researchers want OpenAI to keep seeing and studying how its models reach answers.
Company position
The account does not state OpenAI’s position on this request.
Independent safety examination
Researchers’ calls
The researchers want independent safety groups to examine OpenAI’s AI systems.
Company position
The account does not state OpenAI’s position on outside examinations.
Key facts
- Researchers
- Three former OpenAI researchers
- Employment status
- Recently fired
- Main request
- Keep AI model reasoning visible and available for study
- Additional request
- Allow independent safety groups to examine AI systems
- Stated concern
- Safety risks as AI technology advances










