Reports & Analysis

Explore PaperBench, a benchmark assessing AI's ability to replicate cutting-edge AI research effectively.

PaperBench: Evaluating AI Agents' Replication Skills in Research

Comprehensive review and analysis.

A critical analysis of sycophancy in AI and the changes ahead.

Sycophancy in AI: Findings and Future Improvements

Comprehensive review and analysis.

Explore new findings on how misalignment in language models occurs and methods to reverse it.

Preventing Misalignment in Language Models: New Insights

Comprehensive review and analysis.

Explore the worst-case risks of open weight LLMs with insights on malicious fine-tuning in biology and cybersecurity.

Analyzing Risks of Open Weight LLMs: A Deep Dive

Comprehensive review and analysis.

Explore OpenAI's strategies to combat AI misuse in our in-depth October 2025 report.

Countering Malicious AI Uses: OpenAI's October 2025 Report

Comprehensive review and analysis.

Explore the capabilities and safety evaluations of the gpt-oss-safeguard models.

In-depth Analysis of gpt-oss-safeguard Models

Comprehensive review and analysis.

Explore our report on advanced AI adoption disparities and initiatives for productivity gains.

Addressing Capability Overhang in Global AI Adoption

Comprehensive review and analysis.

In-depth report on GPT-5 adoption trends and its application in various industries.

Exploring GPT-5: Business Use Cases and Trends

Comprehensive review and analysis.

ANALYSIS

State of AI 2025: Key Findings

Comprehensive review and analysis.