Вештачка интелигенција

A Critical Look at the Security Issues with DALL-E 3: Insights from a Microsoft Engineer

Summary

Model artificial intelligence DALL-E 3, developed by startup OpenAI in collaboration with Microsoft, is facing security vulnerabilities that could enable the creation of violent or explicit images, according to Shane Jones, the leader of the engineering team and manager at […]

Model veštačke inteligencije DALL-E 3 ima bezbednosne propuste, tvrdi inženjer Microsofta

Model artificial intelligence DALL-E 3, developed by startup OpenAI in collaboration with Microsoft, is facing security vulnerabilities that could enable the creation of violent or explicit images, according to Shane Jones, the leader of the engineering team and manager at Microsoft. After informing his superiors, Jones attempted to raise awareness through social media, but was allegedly silenced by Microsoft’s legal team.

This issue is reminiscent of the recent case involving Taylor Swift, who became a victim of fake images generated by artificial intelligence that appeared online. As a result, Jones decided to convey his message directly to the US authorities on Capitol Hill.

In his letters to US senators and the Washington Attorney General, Jones outlined his findings regarding the risks that DALL-E 3 poses to public safety, and proposed suspending the public use of the model until these risks are addressed. Despite Microsoft’s suggestion that Jones should directly contact OpenAI, he claims to have discovered that the security vulnerability is associated with DALL-E 3 itself.

Microsoft allegedly requested the deletion of Jones’ post about the vulnerabilities on the social media platform LinkedIn, without providing a detailed explanation. The company asserts its dedication to resolving any issues its employees encounter, but Jones claims otherwise.

Frequently Asked Questions (FAQ) Regarding the Issues with DALL-E 3 Artificial Intelligence: