Google is rolling out new guidelines for app developers who are building AI apps for distribution through the Google Play Store. The new guidelines aim to cut inappropriate and prohibited content. Under these, the company wants apps offering AI features to prevent the generation of restricted content, which includes sexual content, violence, and more.
In addition to this, Google wants app developers to offer a way for users to flag offensive content they find. Google also wants developers to rigorously test their AI tools and models to ensure respect for users' safety and privacy.
Google is also strengthening its guidelines for apps that promote inappropriate use cases, such as those that involve undressing people or creating nonconsensual nude images in their marketing materials. If the app's advertising claims that it can perform such actions, it may be removed from Google Play, regardless of its actual capabilities.
This comes after the surge in AI undressing apps that have been marketing themselves across social media in recent months. An April report by 404 Media, for instance, discovered that Instagram was displaying advertisements for applications that claimed to utilize AI for creating deepfake nudes.
Developers are responsible for protecting their apps against prompts that could manipulate their AI features to create harmful and offensive content. Google advises developers to use its closed testing feature to share early versions of their apps with users for feedback. The company strongly recommends that developers not only test before launching but also document those tests, as Google may request to review them in the future.
Meanwhile, OpenAI has recently announced that it is pausing the use of one of the voices in its AI chatbot, ChatGPT. According to reports, some users have said that the voice sounds similar to Hollywood actress Scarlett Johansson. In a blog post, the AI company, run by Sam Altman, stated that Sky Voice will be paused after being used during the launch of its new GPT-4 model last week.
ALSO READ: What are AI agents and how do they pose a risk to privacy? | Explained
