Moonshot AI faces backlash over Kimi influencer guidelines
A screenshot posted on X appears to show Moonshot AI telling foreign influencers not to label paid Kimi posts or link to the site.
Moonshot AI is facing criticism over allegations that its promotional guidelines for Kimi K3.1 Pro told foreign influencers to conceal the paid nature of their posts, according to aitimes.com.
A screenshot posted on X showed what was described as the company's promotional guidelines for Kimi, asking influencers to submit post drafts for approval and not to add a paid partnership label. The guidelines also reportedly told creators not to include a link to the Kimi website or tag official accounts.
Elshayib, an AI developer who posted the screenshot on X, said he had posted about Kimi himself but did not join the campaign and did not use the provided templates. He criticized the influencers who took part.
The allegations come as Moonshot positions Kimi K3.1 Pro as a model with limited hacking-capable abilities. According to the reports, Kimi K3.1 Pro scored 81.0% on an internal task of finding and exploiting software vulnerabilities in a controlled environment, versus 90.5% for a similar variant without the restrictions. Moonshot said it removed cyberattack-related data from training data and reinforcement learning environments while keeping general coding ability.
Moonshot also said model weights that outside developers can run and modify would be released by November 4.
Quick answers
What do the Kimi promotional guidelines allegedly ask influencers to do?
According to a screenshot posted on X, they asked influencers to submit post drafts for approval, avoid adding a paid partnership label, and not include a link to the Kimi website or tag official accounts.
When will Moonshot release Kimi K3.1 Pro model weights?
Moonshot AI said model weights that outside developers can run and modify would be released by November 4.