OpenAI tries to ‘uncensor’ ChatGPT

Feb 17, 2025

OpenAI revises ChatGPT training for more intellectual freedom.

: OpenAI is altering ChatGPT's training to promote intellectual freedom, allowing for a wider range of responses. The company aims to make the AI less censored by offering multiple perspectives on controversial topics while maintaining a commitment to truth. These changes come amidst political pressures and shifts within Silicon Valley regarding AI and content moderation. However, OpenAI denies these adjustments are appeasing the new Trump administration.

OpenAI has announced a shift in how it trains ChatGPT, aiming to embrace intellectual freedom and neutrality on controversial topics. This involves updating its Model Spec to avoid taking editorial stances and instead presenting various perspectives while maintaining a commitment to truth.

The company has faced criticism from conservatives, including figures like David Sacks and Elon Musk, accusing it of AI censorship. OpenAI's CEO, Sam Altman, has acknowledged some biases, but the recent changes emphasize more control and speech for users.

As part of broader changes in Silicon Valley regarding content moderation, OpenAI removes certain warnings from ChatGPT and follows a trend towards free speech advocated by companies like Meta and X. This comes amid the infrastructural expansion of OpenAI and its goal to challenge Google Search's dominance.