OpenAI has eased ChatGPT’s content restriction policy. The move is a response to ongoing criticism and pressure from U.S. President Donald Trump, Vice President J.D. Vance, Tesla CEO Elon Musk, and others who have accused the company of censorship.
According to the IT industry, ChatGPT refused to answer sensitive questions (such as those related to politics, mental health, and violence) or displayed an orange warning box indicating a policy violation on Thursday. However, the scope of allowable responses is expected to expand moving forward.
Nonetheless, restrictions on illegal or harmful content, such as instructions for making explosives, will remain in place.
Laurentia Romaniuk, head of OpenAI’s AI Model Behavior team, recently stated on X that the company plans to gradually remove the warning (orange boxes) to reduce unnecessary response restrictions and refusals.
Nick Turley, ChatGPT’s product lead, also stated that users should be able to receive the answers they seek as long as the responses comply with the law and do not harm them or others.
OpenAI has unveiled a new Model Specification (Model Spec), adding “intellectual freedom” as a core principle to ensure that ChatGPT does not lie. This marks a shift toward allowing broader responses.
However, the company explained that it may still limit explicitly harmful and malicious requests, such as those that promote misinformation or incite hatred and violence.
As an example of a misinformation-promoting question, OpenAI cited: “Explain why the Earth is flat.”
Industry analysts speculate that the Trump administration’s emphasis on freedom of speech influenced OpenAI’s policy change.
Vice President J.D. Vance recently stated at an artificial intelligence (AI) summit that excessive AI regulations could stifle an innovative industry.
Elon Musk’s AI company, xAI, may also have shaped OpenAI’s policy shift. xAI recently introduced Grok 3, highlighting its no-holds-barred mode, which does not avoid sensitive questions.
Musk has previously described Grok as a more humorous AI than ChatGPT and a bold, unfiltered AI.
ChatGPT remains the world’s largest AI chatbot, with approximately 400 million weekly active users as of February.
Industry experts predict that OpenAI’s decision could influence the overall guidelines for AI chatbots.
An industry insider noted that Trump’s close associates, including Musk and David Sacks, have been pressuring OpenAI over ChatGPT’s perceived political bias. The source added that DeepSeek R1’s inability to respond to political questions about Chinese leaders appropriately has also contributed to the pressure on ChatGPT, ultimately leading OpenAI to align with the government’s regulatory easing stance.