xAI’s promised safety report is MIA

Spread the love

Elon Mask’s AI Organization, Jai, Have missed a self-suffered deadline To publish a final AI protection structure as mentioned by the Watchdog Group The Midas Project.

Jai AI is not known for its anti -security promise because it is usually understood. A recent report It has been found that the company’s AI chatbot, groke, women’s pictures will be dressed. Grock may also be More cres enough Curse without a chattbut, like Jemini and Chatzipt, without much restraint to speak.

Nevertheless, at the AI ​​Seoul Summit in Febru Draft structure Outline the company’s approach to AI protection. The eight -page document kept Jai’s protection priority and philosophy, including the company’s benchmarking protocol and AI model deployment.

As the Midas Project is mentioned on the blog post on Tuesday, but the draft has only applied to the undisclosed future AI models “is not currently in development.” Furthermore, it failed to clearly detect and apply how Jae risk is to detect and apply, it is a key component of a document that is a key component of the organization Signed to the AI ​​COOL SummitThe

In the draft, Jai said that it was planned to publish a revised version of the protection policy “within three months” by May 10. The deadline came and left without recognition on Jai’s official channels.

In spite of frequent alertness of the musk AII Left Did not checkJai has a weak AI protection track record. A recent study by a non -profit Serfi in order to improve the accountability of AI labs shows that Jai is badly located among the peers, because of this “Very weak” risk management practiceThe

It’s not like suggesting other AI labs dramatically better. In recent months, Jai rivals with Google and OpenAI are there The Protection Exam has gone off And has been Slow to reveal Model Protection Report (or Avoid publishing reports Overall). Some experts have expressed concern that the apparent deprivation of security efforts is coming at a time when AI is more capable – and thus the potentially dangerous – than ever.

Leave a Reply

Your email address will not be published.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