Activities
Divisions
Performances
Activities
Divisions
Performances
A Microsoft insider warns US authorities that OpenAI-driven Copilot can effortlessly produce dangerous images
Similar to Google's Gemini AI that generates images, Microsoft's Copilot, fueled by OpenAI's ChatGPT, can also craft highly believable yet deeply troubling false images. An informant alerted US governing bodies and Microsoft's board about this concern.
A Microsoft technician, Shane Jones, has expressed worry over harmful and insulting images created by the firm's AI image-creator tool. Jones, who sees himself as a whistleblower, has appealed to US regulatory bodies and the board of directors at Microsoft, pressing them to intervene, according to an Associated Press report.
Jones recently convened a meeting with the staff of the US Senate to talk about his worries. He also delivered a letter to the Federal Trade Commission (FTC). The FTC acknowledged receiving the letter but chose not to make any more statements.
Microsoft affirmed its dedication to resolving employee worries and valued Jones' work in examining the technology. Nonetheless, it advised utilizing in-house communication methods to probe and rectify the concerns.
Jones, a chief software engineering manager, has dedicated a quarter of a year attempting to tackle safety issues linked to Microsoft’s Copilot Designer. He emphasized the danger of the tool producing damaging material even when triggered with harmless cues. For instance, if prompted with the phrase 'car accident,' Copilot Designer could potentially display unsuitable, sexualized depictions of women.
Jones stressed to FTC Chair Lina Khan that Copilot Designer has the potential to generate inappropriate content, even in response to innocent user inquiries. As an example, he pointed out that it may produce sexualized images of women when asked about 'car accidents.' Furthermore, he underscored issues with other troubling content such as violent themes, political prejudice, underage alcohol consumption and substance abuse, violations of copyright laws, propagation of conspiracy theories, and religious symbols.
Jones has aired these worries before. Initially, Microsoft suggested he go to OpenAI with his discoveries, which he did. In December, he published a message to OpenAI on LinkedIn, resulting in a demand from Microsoft's legal department for its removal. However, Jones has continued to voice his concerns, taking them to the US Senate's Commerce Committee and the office of the Attorney General of Washington State.
Jones emphasized that the primary problem is with OpenAI's DALL-E model, but those using OpenAI's ChatGPT for AI image generation have a lower probability of facing damaging results due to the varying safety precautions taken by the respective companies.
"He communicated through a message that a lot of the issues related to Copilot Designer are already handled by the inherent protective measures of ChatGPT."
The appearance of remarkable AI image creators in 2022, such as OpenAI's DALL-E 2 and the later launch of ChatGPT, sparked a lot of attention from the public. This led to big technology companies like Microsoft and Google to create their own iterations.
Nonetheless, in the absence of strong protective measures, the technology poses threats. It allows users to generate damaging "deepfake" visuals portraying political personalities, war scenarios, or forced nudity, erroneously associating them with actual individuals.
Google put a temporary hold on the Gemini chatbot's ability to generate images due to issues raised regarding racial and ethnic portrayals, especially related to depicting people of color in uniforms from the Nazi-era.
(Incorporating information from various sources)
Search for us on YouTube
Highlighted Programs
Connected Narratives
NVIDIA's Jensen Huang asserts that AI hallucinations can be resolved, and anticipates that artificial general intelligence will be a reality in approximately 5 years.
OpenAI's Sora has the capability to produce lifelike nude videos, with developers swiftly working towards implementing a solution.
Apple has at last introduced MM1, its own multimodal AI model designed to generate text and images.
Microsoft has brought on board Mustafa Suleyman, the cofounder of DeepMind, to head their new consumer AI division.
You can find more on YouTube.
Firstpost holds all rights reserved, copyright 2024.
+ There are no comments
Add yours