Brother
Professional
- Messages
- 2,590
- Reaction score
- 498
- Points
- 83
Problems in DALL-E 3 have raised fears of Shane Jones abusing the technology.
The network is actively discussing the scandalous statement of a Microsoft manager about flaws in the DALL-E 3 security system from OpenAI. According to Shane Jones, the technology allows you to create violent and obscene images. He claims to have discovered a way to bypass the DALL-E 3 security systems in December 2023 and reported it to his management at Microsoft.
Jones paid particular attention to the recent Taylor Swift deepfake porn scandal, which the engineer called an example of potential abuse of technology. Jones claims that vulnerabilities in DALL-E 3 and products that use it, such as Microsoft Designer, make it easier to abuse AI when creating malicious images.
The engineer called on OpenAI to temporarily withdraw DALL-E 3 from the market and reported his concerns to Microsoft. Jones claims that the company prevented him from drawing public attention to the problem. The engineer posted a bug letter on LinkedIn sent to US senators and the Washington State Attorney General, but Microsoft demanded that it be removed.
A Microsoft representative stressed that the company listens to the comments of employees and conducted its own investigation, confirming that the techniques reported by Jones did not bypass the company's security systems. In addition, Microsoft has an internal tool for reporting problems related to artificial intelligence.
Jones said that his experience shows the need for an independent and effective reporting system. He used Microsoft's internal mechanisms, but in response received a recommendation to contact the OpenAI website directly.
An OpenAI spokesperson said that the company immediately investigated the message from a Microsoft employee and confirmed that the detected method does not bypass their security systems. It is noted that DALL-E 3 introduced filters to prevent the creation of malicious images, as well as additional security measures in ChatGPT products and the DALL-E API.
Jones calls on the government to create a system for reporting risks and problems related to AI, and ensure that employees can use such a system without fear of possible reprisals. A responsible approach to AI security is a key aspect in the development of the field, and the situation with DALL-E 3 highlights the importance of such an approach for ensuring the safety and ethics of using artificial intelligence.
The network is actively discussing the scandalous statement of a Microsoft manager about flaws in the DALL-E 3 security system from OpenAI. According to Shane Jones, the technology allows you to create violent and obscene images. He claims to have discovered a way to bypass the DALL-E 3 security systems in December 2023 and reported it to his management at Microsoft.
Jones paid particular attention to the recent Taylor Swift deepfake porn scandal, which the engineer called an example of potential abuse of technology. Jones claims that vulnerabilities in DALL-E 3 and products that use it, such as Microsoft Designer, make it easier to abuse AI when creating malicious images.
The engineer called on OpenAI to temporarily withdraw DALL-E 3 from the market and reported his concerns to Microsoft. Jones claims that the company prevented him from drawing public attention to the problem. The engineer posted a bug letter on LinkedIn sent to US senators and the Washington State Attorney General, but Microsoft demanded that it be removed.
A Microsoft representative stressed that the company listens to the comments of employees and conducted its own investigation, confirming that the techniques reported by Jones did not bypass the company's security systems. In addition, Microsoft has an internal tool for reporting problems related to artificial intelligence.
Jones said that his experience shows the need for an independent and effective reporting system. He used Microsoft's internal mechanisms, but in response received a recommendation to contact the OpenAI website directly.
An OpenAI spokesperson said that the company immediately investigated the message from a Microsoft employee and confirmed that the detected method does not bypass their security systems. It is noted that DALL-E 3 introduced filters to prevent the creation of malicious images, as well as additional security measures in ChatGPT products and the DALL-E API.
Jones calls on the government to create a system for reporting risks and problems related to AI, and ensure that employees can use such a system without fear of possible reprisals. A responsible approach to AI security is a key aspect in the development of the field, and the situation with DALL-E 3 highlights the importance of such an approach for ensuring the safety and ethics of using artificial intelligence.
