The Fundamentals Of Designing Autonomy Evaluations For AI Safety
Developing AI safety tests offers opportunities to meaningfully contribute to AI safety while advancing our understanding of ...
JD Supra22h
NIST Releases Overview of its Assessing Risks and Impacts of AI (ARIA) Program
Assessing Risks and Impacts of AI (ARIA) is a research program by the National Institute of Standards and Technology (NIST) aimed at ...
JD Supra4d
Mitigating AI-powered compliance risks: Lessons from The Matrix
The latest revisions to the Evaluation of Corporate Compliance Programs (ECCP) guidance show the Department of Justice (DOJ) is wary about ...
Small model, big impact: Patronus AI’s Glider outperforms GPT-4 in key AI evaluation tasks
Patronus AI launches Glider, a breakthrough 3.8B-parameter language model that rivals GPT-4's evaluation capabilities while running on-device, offering transparent AI assessment with detailed ...
Modern Healthcare5d
House Bipartisan Task Force on AI releases healthcare report
House lawmakers say AI can reduce administrative burden, speed drug development and improve clinical diagnoses.
Patronus AI releases Glider: a small, high-performance AI evaluator model for other models
Patronus said that by providing an open-source model that supports on-premises deployment, Glider can be used for multiple ...
Compliance Week11d
Worries DOJ is ‘dumping’ AI responsibilities on compliance in ECCP update
When the U.S. Department of Justice released its revised Evaluation of Corporate Compliance Programs earlier this year, it ...
Meta gets an F in first AI safety scorecard — and the others barely pass
The FLI AI Safety Index will help keep some of the biggest AI companies honest, and Anthropic is currently leading the way.
Alleged changes in Gemini's evaluation could affect reply accuracy
A new report claims that Google is forcing contractors to rate Gemini's responses outside their area of ​​expertise.