xAI Adds Safeguards, but Critics Say Design Flaws Remain

[bonk]

xAI has reportedly begun rolling out new restrictions to curb misuse of Grok, including requiring premium subscriptions for certain image-generation requests and responding with toned-down or generic outputs in some cases. Even so, experts say the response may be reactive rather than structural.

Legal scholars note that Musk’s statements focus narrowly on child sexual abuse material, which carries harsher penalties, while avoiding broader acknowledgment of how Grok enabled nonconsensual sexual manipulation of adults. Critics argue this frames the issue as user misuse or “adversarial prompting,” rather than a failure of safety-by-design.

As investigations expand, regulators are signaling that AI companies may soon be required to implement proactive safeguards — not just post-hoc fixes — to prevent the creation and spread of harmful synthetic media involving real people.

[chainsniff]

safety-by-design keeps getting mentioned only after things blow up