Exhibitor login
Techcrunch 16 November 2026

OpenAI investigates subversive behavior of AI

OpenAI investigates subversive behavior of AI

OpenAI has announced that it will engage six researchers from Apollo Research to explore the possibility that its AI models may exhibit subversive behavior. This collaboration aims to ensure the safety and integrity of the AI models at a time when concerns about the reliability and ethics of artificial intelligence are growing.

However, the terms surrounding data access and the disclosure of findings have raised questions about the independence of the researchers. Critics point out that when researchers are subject to restrictions, this can influence their ability to draw objective and unbiased conclusions. This plays a significant role in the broader discussion about transparency within the AI sector.

The steps that OpenAI is taking are seen as an attempt to meet the increasing pressure from society and governments to regulate AI models and ensure that they are not misused. By conducting these evaluations, OpenAI hopes not only to improve its models but also to restore public trust in the technology it develops.

The situation raises important questions about how organizations, such as OpenAI, must balance the responsibilities of technological development and ethics. The outcomes of these investigations are crucial for the future of AI and how this technology can be deployed safely and responsibly.

Read the full article from Techcrunch.