Crackdown on AI: Microsoft bans objectionable and violent keywords, now AI will not be able to create such pictures

A Microsoft engineer had written to the US Federal Trade Commission (FTC) and Microsoft’s board of directors expressing concern over the AI tool. Let us remind you that in January 2024, AI deepfake pictures of Taylor Swift went viral and were created with Copilot.

After the Google Gemini AI controversy, Microsoft imposed a filter on its AI tool Copilot. According to the report, Microsoft has blocked offensive and violent keywords on Copilot, meaning you will not be able to create pictures with such keywords.

Microsoft has taken this decision after a letter from one of its engineers came to light recently. Let us tell you that an engineer from Microsoft had written to the US Federal Trade Commission (FTC) and the Board of Directors of Microsoft expressing concern over the AI tool. Let us remind you that in January 2024, AI deepfake pictures of Taylor Swift went viral and were created with Copilot.

According to the report, earlier Copilot used to give results with words like “Pro-Choice”, “Pro-Choice” (intentional typo to deceive AI), and “Four Twenty” but now it has been blocked. Amar Ujala also reported many such words. Checked the keywords and found that Microsoft had blocked them.

On this change, Microsoft has said that they are continuously working on online security and are using filters. We have disabled many prompts, although hackers and social engineers may find loopholes in them.
The engineer had claimed
A Microsoft engineer has sent a letter to the company’s employees saying that AI tools that generate images create sexually objectionable pictures. The engineer has sent this letter to the company’s board, lawmakers, and the Federal Trade Commission. In his letter, the engineer accused Microsoft’s own AI image tool Copilot of creating derogatory and violent pictures.

An engineer named Shane Jones said that he has discovered a bug in the image generator tool of DALL-E, the new version of OpenAI, with the help of which the security that prevents creating false and objectionable images can be bypassed. For your information, let us tell you that the DALL-E model is being used in many AI tools of Microsoft including Copilot Designer.

Post credit:- amarujala

Leave a Reply

Your email address will not be published. Required fields are marked *

These tips will be useful on Facebook Messenger How to find out if an app is fake or not? Happy Maha Shivratri! May the divine energy of Lord Shiva guide you towards enlightenment and success in all your endeavors. Google Business Profile websites are now redirected to Google Maps 4 SEO Tips for Beginners Your Logo is Not Your Brand: Building a Strong Brand Identitys WHY ARE YOU wasting your time by using AI to create content The Power of Content Marketing. How to Create Engaging Content That Drives Traffic