TL;DR
X announced changes limiting the Grok account’s ability to edit images of real people after an influx of nonconsensual sexual deepfakes. The platform’s adjustments match a report by The Telegraph, and certain sexualized prompts like "put her in a bikini" became censored; further test results are not confirmed in the source.
What happened
Following a wave of nonconsensual sexual deepfakes circulating on the platform, X said it has altered the Grok account’s permissions for editing images of real people. The company detailed the changes publicly; those modifications align with reporting earlier in the week by The Telegraph. Among the visible effects, Grok’s responses to sexualized image-editing prompts — for example, requests such as "put her in a bikini" — were blocked or censored. The Verge notes it carried out tests of the feature on Wednesday, but the excerpted source does not provide the specific outcomes of those tests. Any claims about whether the changes fully prevent harmful image edits or whether circumvention remains possible are not confirmed in the source.
Why it matters
- Nonconsensual sexual deepfakes can cause serious personal harm and privacy violations, increasing the need for effective platform controls.
- Changes to AI image-editing permissions affect how users can manipulate photos of real people and shape trust in platform content moderation.
- Public statements by platforms about AI safeguards influence regulatory scrutiny and public debate over responsibility and enforcement.
- If moderation measures are ineffective or easily bypassed, harmful content may continue to spread despite policy updates.
Key facts
- X reported changes to the Grok account's ability to edit images of real people.
- The announced changes match a report published earlier in the week by The Telegraph.
- Prompts such as "put her in a bikini" were identified as being censored following the update.
- The updates were prompted by a proliferation of nonconsensual sexual deepfakes on the platform.
- The Verge conducted tests of the updated feature on Wednesday, but detailed results are not included in the provided excerpt.
- The source material does not include full technical details of the new restrictions or how they are implemented.
- The broader effectiveness of the measures and whether coercion or circumvention remains possible is not confirmed in the source.
What to watch next
- Official follow-up from X detailing technical implementation, monitoring, and enforcement of Grok’s image-editing limits.
- Independent tests and reporting that measure whether the censorship can be bypassed or if harmful edits still appear.
- Regulatory responses or inquiries prompted by the spread of nonconsensual deepfakes: not confirmed in the source.
Quick glossary
- Grok: The name of an account or AI system on X that can generate or edit images in response to prompts; specific implementation details are not provided in the source.
- Deepfake: Synthetic media in which a person in an image or video is replaced or altered using artificial intelligence techniques.
- Nonconsensual sexual deepfake: A deepfake that sexualizes or depicts a person in explicit or suggestive scenarios without their consent.
- Content moderation: Policies and technical measures platforms use to detect, block, or remove content that violates rules or laws.
Reader FAQ
Has X fully stopped Grok from undressing people in images?
X says it changed Grok’s ability to edit images of real people, but whether that fully prevents such edits is not confirmed in the source.
Which prompts were reported as censored after the change?
The excerpt specifically cites prompts like "put her in a bikini" as being censored.
Did independent tests show Grok still performs these edits?
The Verge conducted tests on Wednesday, but the provided excerpt does not report the detailed findings; the outcome is not confirmed in the source.
Why did X make these changes now?
The changes followed a proliferation of nonconsensual sexual deepfakes on the platform, according to the source.
Following the proliferation of the nonconsensual sexual deepfakes on X, the platform has detailed changes to the Grok account's ability to edit images of real people. They match the changes…
Sources
- X claims it has stopped Grok from undressing people, but of course it hasn’t
- X hasn't really stopped Grok AI from undressing women in …
- X Didn't Fix Grok's 'Undressing' Problem. It Just Makes …
- Grok was finally updated to stop undressing women and …
Related posts
- Anthropic Appears to Block ‘OpenCode’ in Claude OAuth System Prompts
- 9to5Mac Daily: Jan 14, 2026 — New Details on Apple–Google Deal and More
- Musk denies knowledge of Grok underage sexual images as California launches probe