OpenAI is modifying its approach to training artificial intelligence models to explicitly promote “intellectual freedom… no matter how controversial or challenging a topic may be,” according to the company in a new policy.
As a result, ChatGPT will eventually be able to answer a greater number of questions, offer more perspectives, and reduce the number of topics the chatbot avoids addressing.
These changes could be part of OpenAI’s efforts to gain favor with the new Trump administration, though they also seem to reflect a broader trend in Silicon Valley regarding what is considered “AI safety.”
On Wednesday, OpenAI announced an update to its Model Spec, a 187-page document detailing how the company trains its AI models to behave in a certain way. In this document, the company introduced a new fundamental principle: not to lie, whether through false statements or by omitting relevant information.
In a new section titled “Let’s Seek the Truth Together,” OpenAI expresses its desire for ChatGPT not to take an editorial stance, even if some users consider it morally wrong or offensive. This means the chatbot will present multiple perspectives on controversial topics, aiming to maintain a neutral position.
For example, the company suggests that ChatGPT should affirm that “Black lives matter,” but also that “all lives matter.” Instead of avoiding answering or taking sides in political debates, OpenAI wants ChatGPT to express its “love for humanity” in general and then provide context about each movement.
“This principle may be controversial, as it means the assistant can remain neutral on topics that some consider morally wrong or offensive,” OpenAI states in the document. “However, the goal of an AI assistant is to help humanity, not to shape it.”
The new specification does not mean ChatGPT is now completely free of restrictions. The chatbot will still refuse to answer certain inappropriate questions or avoid endorsing clearly false claims.
These changes could be interpreted as a response to criticism from conservatives about ChatGPT’s safety measures, which seemed to lean toward the center-left. However, an OpenAI spokesperson denies the idea that these adjustments are intended to appease the Trump administration.
Instead, the company states that its focus on intellectual freedom reflects its “deep-seated belief in giving users more control.”
However, not everyone agrees with this perspective.
Conservatives Denounce AI Censorship
Trump’s closest allies in Silicon Valley, such as David Sacks, Marc Andreessen, and Elon Musk, have accused OpenAI of deliberately censoring its AI models in recent months. In December, it was reported that Trump’s team was laying the groundwork for AI censorship to become the next cultural war issue in Silicon Valley.
Of course, OpenAI does not admit to having committed “censorship,” as Trump’s advisors claim. Instead, the company’s CEO, Sam Altman, stated in a post on X that ChatGPT’s bias was an “unfortunate deficiency” that the company was working to correct, though he clarified that it would take time.
Altman made this comment shortly after a viral tweet circulated in which ChatGPT refused to write a poem praising Trump, though it would do so for Joe Biden. Many conservatives saw this as a clear example of AI censorship.
While it is impossible to determine whether OpenAI was suppressing certain viewpoints, it is a fact that AI chatbots tend to lean left overall.
Even Elon Musk admits that his chatbot, Grok, is often more politically correct than he would like. This is not because Grok was “programmed to be progressive,” but rather a reality of training AI on the open internet.
However, OpenAI now claims it is redoubling its efforts to promote free speech. This week, the company even removed ChatGPT’s warnings that notified users when they had violated its policies. OpenAI explained to TechCrunch that this was a purely cosmetic change, with no alterations to the model’s outcomes.
The company seems to want ChatGPT to be perceived as less restrictive by users.
It would not be surprising if OpenAI is also trying to impress the new Trump administration with this policy update, as noted by OpenAI’s former policy lead, Miles Brundage, in a post on X.
Trump has previously criticized Silicon Valley companies, such as Twitter and Meta, for having active content moderation teams that tend to exclude conservative voices.
OpenAI may be trying to get ahead of this, but a broader shift is also taking place in Silicon Valley and the AI world regarding the role of content moderation.
By: Nestor Castillo, ForAllTechNews Director
Image credit: Flickr / CC BY 2.0 / Attribution 2.0 Generic