The weekend is almost here but before you zone out, here are the most noteworthy developments from the artificial intelligence world. In the first incident, Discord, the popular social media platform is getting rid of its in-house experimental AI chatbot Clyde. It will no longer be available from December 1, 2023. In other news, Microsoft has made changes to its AI image generator tool after it created very close resembling images of Disney posters, along with its logo. This and more in today’s AI roundup. Let us take a closer look.
Discord is shutting down its AI chatbot
Discord is discontinuing Clyde, its experimental AI chatbot, by deactivating it at the end of the month, as per a note by the company. Users will no longer be able to invoke Clyde in direct messages, group messages, or server chats starting from December 1st. The chatbot, which leveraged OpenAI’s models for answering questions and engaging in conversations, had been in limited testing since earlier in the year, with initial plans to integrate it as a fundamental component of Discord’s chat and communities app.
Microsoft tweaks its AI image generator
Microsoft has adjusted its AI image generator tool following concerns over a social media trend where users used the tool to create realistic Disney film posters featuring their pets, reports the Financial Times. The generated images, posted on TikTok and Instagram, raised copyright issues as Disney’s logo was visible. In response, Microsoft blocked the term “Disney” from the image generator, displaying a message stating the prompt was against its policies. It is suggested that Disney may have reported concerns related to copyright or intellectual property infringement.
PM Modi highlights the problem of deepfakes
Prime Minister Narendra Modi highlighted the increasing problem of deepfakes in India while addressing journalists at the Diwali Milan program at the BJP headquarters in New Delhi. ‘I watched my deep fake video in which I’m doing Garba. But the reality is that I have not done garba after my school life. Someone made my deepfake video”, said PM Modi.
ANI also quoted him as saying, “There is a challenge arising because of Artificial Intelligence and deepfake…a big section of our country has no parallel option for verification…people often end up believing in deepfakes and this will go into a direction of a big challenge…we need to educate people with our programs about Artificial Intelligence and deepfakes, how it works, what it can do, what all challenges it can bring and whatever can be made out of it”.
Senior Stability AI executive resigns over copyright issues
A senior executive, Ed Newton-Rex, has resigned from the AI-focused company Stability AI due to the company’s stance that using copyrighted work without permission for training its products is acceptable. Newton-Rex, former head of audio at the UK and US-based company, told BBC that he deemed such practices as “exploitative” and against his principles. However, many AI firms, including Stability AI, argue that using copyrighted content falls under the “fair use” exemption, which allows for the use of copyrighted material without obtaining permission from the original owners.
Research finds popular AI image generators can be tricked
Researchers successfully manipulated Stability AI’s Stable Diffusion and OpenAI’s DALL-E 2 text-to-image models to generate images in violation of their policies, including depictions of nudity, dismembered bodies, and violent or sexual scenarios. The study, set to be presented at the IEEE Symposium on Security and Privacy in May, highlights the vulnerability of generative AI models to bypass their own safeguards and policies, a phenomenon referred to as “jailbreaking.” This research underscores the challenges in ensuring responsible and ethical use of AI technologies. A preprint version of the study is available to see on arXiv.