TECH NEWS – ChatGPT creators consider allowing erotic content but say it won’t make it easier to create deepfakes…

OpenAI really took off after the public presentation of GPT-4o. This is ChatGPT’s next step in bringing artificial intelligence closer to human capabilities. After Sam Altman (the company’s CEO) promised that we would see “something magical,” the event lived up to expectations and impressed many with the capabilities of OpenAI’s bot.

However, shortly before the presentation, the company spoke publicly about a possible policy change that could lead to very dangerous situations.

As reported in a recent publication of the TechSpot portal, OpenAI is investigating the possibility that users can use artificial intelligence to create porn films. Currently, artificial intelligence companies use filters to prevent them from making such content. However, the US company is considering revising its terms. All this is done in order to allow users to responsibly create erotic content with tools that use generative artificial intelligence.

Deepfakes can be a major concern for ChatGPT plans

However, this possibility raises the issue of the proliferation of a problem that has already affected celebrities such as Taylor Swift: unauthorized deepfakes. With the help of tools based on artificial intelligence, many users created pornographic content about the artist. It spreads unchecked on social networks.

Joanne Jang, the current head of the OpenAI model, said that they will be very careful with this decision as they don’t want people to violate other people’s rights.

Jang talked about the possibility of removing their filters. He also touched on what kind of content can be created: “It depends on the definition of pornography.” However, he was definite about deepfakes. He stated that under no circumstances will they allow this type of content to be created. Be that as it may, the existence of this possibility is widely feared. Because removing these filters and giving so much freedom to an AI can be counterproductive to producing a safe and valuable language model.

Source: TechSpot