Activities
Divisions
Performances
Activities
Divisions
Performances
A person revealing inside information from Microsoft has informed US authorities that OpenAI's Copilot can generate damaging pictures 'too effortlessly'. Similar to Google's Gemini AI image generator, Microsoft's Copilot, running on OpenAI's ChatGPT, can also produce some extremely troublesome yet highly believable counterfeit images. This insider alerted US oversight bodies and Microsoft's board about the potential problem by sending them letters.
A Microsoft employee has expressed worry over inappropriate and damaging visuals created by the firm's AI image-creation software. Shane Jones, who views himself as a whistleblower, has written to American authorities and Microsoft's board, advocating for them to intervene, according to an Associated Press report.
Jones had a recent meeting with personnel from the US Senate to express his worries, and he also wrote a letter to the Federal Trade Commission (FTC). The FTC acknowledged receiving the letter, but chose not to provide any additional information.
Microsoft expressed its dedication to tackling staff issues and acknowledged Jones' contributions toward examining the technology. Nonetheless, it advised employing in-house communication pathways to scrutinize and resolve the problems.
Jones, a chief software engineering supervisor, has dedicated the past quarter to tackling safety issues regarding Microsoft's Copilot Designer. He emphasized the potential hazard of the tool producing detrimental content even with harmless cues. For instance, if the phrase 'car accident' is used as a prompt, Copilot Designer could generate unsuitable, sexualized depictions of women.
Jones highlighted to FTC Chair Lina Khan that Copilot Designer can inadvertently produce detrimental material even when prompted by harmless user inputs. For example, if given the prompt 'car accident,' it occasionally generates sexually suggestive images of women. Jones also brought attention to other problematic content such as depictions of violence, political prejudice, underage alcohol consumption and drug usage, violations of copyright laws, propagation of conspiracy theories, and religious symbols.
Jones has voiced his worries in the past. Initially, Microsoft recommended he take his discoveries to OpenAI, which he did. In December, he shared a letter to OpenAI on LinkedIn, causing Microsoft's legal department to insist on its removal. Nevertheless, Jones continued to press on, presenting his issues to the US Senate’s Commerce Committee and the Washington State Attorney General's office.
Jones emphasized that although the primary concern is with OpenAI's DALL-E model, those using OpenAI's ChatGPT for AI image creation are less prone to coming across damaging results. This is because the two companies have put different safety precautions in place.
"He communicated through text that most of the issues related to Copilot Designer are already addressed by the inherent security measures of ChatGPT."
The appearance of remarkable AI image generators in 2022, such as OpenAI's DALL-E 2 and the later launch of ChatGPT, sparked considerable public curiosity. This led major technology companies like Microsoft and Google to create their own variants.
Nonetheless, if strong protective measures aren't in place, the technology poses dangers. It allows users to generate damaging "deepfake" pictures showcasing political leaders, battle scenarios, or unauthorized nudity, wrongfully associating them with actual people.
Google has temporarily disabled the image creation function of the Gemini chatbot due to some issues. These problems primarily revolve around the sensitive topic of racial and ethnic representation, such as examples where individuals of color are portrayed in uniforms from the Nazi era.
(Incorporating information from various sources)
Search for us on YouTube
Highlighted Programs
Associated Articles
NVIDIA’s Jensen Huang believes AI hallucinations can be resolved and predicts artificial general intelligence will be a reality in approximately 5 years.
OpenAI's Sora has the ability to create authentic appearing nude videos, forcing developers to hastily implement a solution.
Apple has finally unveiled its MM1, a multimodal AI model designed for text and image creation.
Microsoft has recruited Mustafa Suleyman, cofounder of DeepMind, to head their new consumer-focused AI division.
You can find more on YouTube.
Firstpost holds all rights and protections, as recognized by copyright law, as
+ There are no comments
Add yours