Discord Cancels Experimental AI Chatbot Clyde, Microsoft Adjusts Image Generator for Copyright Issues
Discord, the popular social media platform, has announced that it will be shutting down its in-house experimental AI chatbot named Clyde. According to a note from the company, Clyde will be deactivated at the end of the month, and users will no longer be able to interact with it starting from December 1st, 2023. This decision comes after limited testing of the chatbot, which utilized OpenAI’s models for answering questions and engaging in conversations. Initially, Clyde was intended to be integrated as an essential component of Discord’s chat and communities app.
In other news, Microsoft has made adjustments to its AI image generator tool in response to concerns over copyright issues. The tool had been used by users to create realistic Disney film posters featuring their pets, which raised copyright concerns as Disney’s logo was visible in the generated images. To address this, Microsoft has now blocked the term Disney from the image generator, displaying a message stating that the prompt violates its policies. It is speculated that Disney may have reported concerns regarding copyright or intellectual property infringement.
Deepfake technology has also caught the attention of India’s Prime Minister Narendra Modi, who highlighted the increasing problem of deepfakes during a recent address to journalists. He revealed that he had seen a deepfake video of himself doing a traditional dance called Garba, despite not having performed the dance since his school days. The prime minister emphasized the challenge posed by deepfakes and artificial intelligence, especially since a significant portion of the population lacks alternative means for verification. He stressed the need for educational programs to raise awareness about artificial intelligence and deepfakes, their capabilities, potential challenges, and the implications they can have.
Meanwhile, Senior Executive Ed Newton-Rex has stepped down from the AI-focused company Stability AI due to a disagreement over its approach to using copyrighted content without permission for training purposes. Newton-Rex, who previously served as the head of audio at the UK and US-based firm, deemed such practices as exploitative and against his principles. However, AI companies like Stability AI argue that using copyrighted material falls under the fair use exemption, allowing for its use without obtaining permission from the original creators.
Research has also shed light on the vulnerability of generative AI models to being manipulated in violation of their own policies. Both Stability AI’s Stable Diffusion and OpenAI’s DALL-E 2 text-to-image models were successfully exploited to generate images depicting nudity, dismembered bodies, and violent or sexual scenarios. Referred to as jailbreaking, this phenomenon demonstrates the challenges in ensuring responsible and ethical use of AI technologies. The study, set to be presented at the upcoming IEEE Symposium on Security and Privacy, highlights the importance of addressing these vulnerabilities.
As the AI landscape continues to evolve, companies are confronting numerous challenges related to ethical use, copyright infringement, and the potential for misuse. It is crucial for industry players to prioritize responsible AI development and deployment to safeguard against unintended consequences while harnessing the technology’s potential for positive impact.
Overall, these recent developments in the AI realm demonstrate the ongoing efforts to navigate the complex landscape of artificial intelligence, ensuring compliance with legal and ethical standards while fostering innovation.
Read more: [Discord Cancels Experimental AI Chatbot Clyde | Microsoft Adjusts Image Generator for Copyright Issues | AI Roundup]