Is Grok 4.7 Content Detectable? Testing SpaceXAI’s Latest AI Model

On September 21st, SpaceXAI announced Grok 4.7, the latest model in its Grok series. 

SpaceXAI (rebranded from xAI) describes Grok 4.7 as its “...most powerful model for coding and knowledge work.”

The arrival of Grok 4.7 comes only about a month after SpaceXAI released Grok 4.6 in early August.

Recently, the race to build and launch the latest AI models has been heating up. Just consider these launches from leading AI companies:

‍

So, in the context of all the rapid releases of new AI models, is Grok 4.7 content still detectable by Originality.ai? We ran an analysis and share our findings in this study!

The Quick Answer — Is Grok 4.7 Text Detectable?

Yes! Originality.ai detected 100% of the Grok 4.7 responses in this study using AI Allowance at 15% settings: 1,000 out of 1,000 English samples.

The Grok 4.7 Detectability Study

New LLM releases make ongoing testing of AI detector accuracy essential. 

At Originality.ai, we test recently released models, as they are launched, to see how our AI detector performs on their generated content.

For this study, we generated writing with Grok 4.7 and scanned the responses with Originality.ai AI Allowance. 

Here’s how we ran the test.

The Grok 4.7 Detectability Test

The Dataset:

We used a fixed set of 1,000 English prompts from our earlier model benchmarks, covering:

• Content types: blogs, conversational text, expository articles, news and reviews, with 200 prompts in each category.
• Prompt styles: requests for a human writing style, rewrites, and minor imperfections.
• Response lengths: 151–1,749 words, with a median of 338 words.

Grok 4.7 generated one final response per prompt using low reasoning effort. We scanned each complete response without editing it. 

Four unusable prompts were replaced within their content categories before any detector scores were reviewed.

All 1,000 final responses were successfully scanned. We classified a response as Likely AI when its AI Allowance confidence score was at least 0.50.

The Grok 4.7 Detectability Results

Originality.ai identified every response in the test set as Likely AI. Here is the overall result:

‍

The detection rate was 100% in all five content categories, with 200 of 200 responses detected in each. 

It also remained at 100% for the longer responses: 

• 929 of 929 at 200+ words
• 612 of 612 at 300+ words

The 996 responses from unchanged prompts were all detected as well, so excluding the four replacements did not change the result.

Final Thoughts

Grok 4.7 content was highly detectable in our test. Originality.ai AI Allowance at 15% settings identified all 1,000 English responses across five content categories.

So yes, Grok 4.7 is still highly detectable by Originality.ai!

Whether you’re a writer, editor, student, or teacher, Originality.ai is an exceptionally effective way to identify whether content is generated and Likely AI or Original. Try Originality.ai’s AI detector.

More on AI detection:

• Originality.ai AI Detection Accuracy Study
• Third-Party AI Detection Studies: Meta-Analysis

Explore our other model detectability studies:

• Is Grok 4.6 content detectable?
• Is Gemini 3.7 Flash content detectable?
• Is Claude Fable 5 content detectable?

‍Methodology Note: This study measures detection of known AI text. It did not include human-written controls or measure false positives, and the result does not guarantee detection of every Grok response.