How it scores

Usefulness
4/5

Tom's Guide: Grok 4.1 narrowly beat ChatGPT-5.2 (tomsguide.com/ai/i-tested-chatgpt-5-2-vs-grok-4-1…); TechRadar: top maths score in OmniCalculator tests; individual reviewers report heavy moderation blocking basic requests.

Ease of use
4/5

Free apps on iOS and Android with 4.83 from 144,419 UK App Store ratings (apps.apple.com/gb/app/grok-ai-assistant…), but limits and modes confuse paying users.

Value for money
2/5

UK App Store lists SuperGrok at £30, above the £8-£20 peer range, and user threads describe cancellations over weekly limits and price rises.

Privacy and trust
1/5

Open Ofcom, ICO, European Commission and Irish DPC investigations over sexualised deepfakes (ofcom.org.uk, theregister.com 2026/02/17); 370,000 shared chats indexed (fortune.com); Common Sense Media rates it unacceptable with unenforced age checks.

Momentum in 2026
4/5

a16z 7th edition web #6 and mobile #19 (a16z.com/100-gen-ai-apps-7/); $37.4M App Store and Play revenue in Aug 2026 (appfigures.com); but its own user forums report users cancelling in Sep-Oct 2026.

Verdict

Grok scores well in head-to-head tests, winning Tom's Guide's December 2025 face-off with ChatGPT and topping one maths ranking reported by TechRadar in May 2026. In January 2026 Ofcom, the ICO, the European Commission and Ireland's data regulator all opened investigations after Grok was used to make sexualised images of real people, including children, and none had concluded by October. Common Sense Media rates it unacceptable for under-18s and its UK App Store paid tier SuperGrok lists at £30.

Privacy note. Open Ofcom, ICO, European Commission and Irish DPC investigations over sexualised deepfakes (ofcom.org.uk, theregister.com 2026/02/17); 370,000 shared chats indexed (fortune.com); Common Sense Media rates it unacceptable with unenforced age checks.

Best for

Adults who already use X and want answers that draw on posts there

Skip if

Its record on sexualised deepfakes and the open regulator investigations are dealbreakers for you

Key facts

Made by
xAI (SpaceXAI, part of SpaceX; UK App Store seller Twitter International Unlimited Company)
Based in
United States
Launched
2025
Works on
iPhone, Android, Web
In the UK
Yes
Price from
£10.0 a month
Free plan
Free plan with daily or weekly usage limits; image generation was moved to paid tiers in January 2026 (Common Sense Media)

Pricing

PlanPriceBilled
SuperGrok Lite£10month
SuperGrok£30month
SuperGrok Plus£100month
SuperGrok Heavy£300month

source source

What it does

  • chats and answers questions with xAI's Grok models
  • creates AI images and videos with Grok Imagine
  • holds spoken conversations in Voice mode
  • builds and deploys simple web apps from a single prompt
  • offers coding tools

Ratings

What people say

Some paying users rate Grok's tone and accuracy above rivals, but Grok's own user forums in autumn 2026 are full of cancellations over tighter moderation, weekly limits and price rises, and complaints that newer versions lost personality. Store stars (4.83 from 144,419 UK App Store ratings) sit far above Trustpilot (1.4 from 519) and user reviews, a big gap.

“I have premium subscription of each of ChatGPT, Claude, Gemini, SuperGrok and Perplexity… For the tone and accuracy, Grok is superior than all others. I enjoy it the most.”

Individual user review, 4 September 2026 · source

“It’s getting worse day by day, the moderations are regular and the worst part is the limits get deducted even if the video gets moderated.”

Individual user review, 6 October 2026 · source

“It was worth every penny until 4.3. Now I'm not so sure.”

Individual user review, 4 September 2026 · source

“The chats are not even generating basic things …saying its moderated”

Individual user review, 6 October 2026 · source

“Already switched back. Main reasons: 1) bot load lag 2) few times I’ve fired a prompt in Grok app, nothing happens…”

Post on X, 26 September 2026 · source

Press reviews

  • Tom's Guide 2025-12-16

    Grok 4.1 'just about took the win' over ChatGPT-5.2 on detail and infographics, but struggled more on logic and creative writing.

  • TechRadar 2026-05-01

    Reports OmniCalculator testing that found Grok 4.2 the strongest free model at maths and logic, while Claude led on writing.

  • Common Sense Media 2026-04-13

    Rated 'Unacceptable' risk; says Grok is far too risky for teens and recommends no use under 18 (date is the latest screenshot on the page).

Privacy

Trains on your data
yes by default for consumer chats, per secondary sources (Anonyome); xAI's privacy policy returned 403 to every fetch, so the official wording is unverified
Opt-out
Settings > Data Controls > untick 'Improve the model' in the app (Settings > Data on grok.com); Private Chat mode is not used for training (Anonyome, secondary)
Retention
no official retention period read
Incidents
More than 370,000 shared Grok chats were indexed by search engines (Forbes via Fortune, 22 Aug 2025). After Grok was used to make sexualised images of real people including children, Ofcom opened an Online Safety Act investigation into X on 12 Jan 2026 and is separately assessing xAI (ongoing, no decision found); the ICO opened formal investigations into X and xAI on 4 Feb 2026 (ongoing); the European Commission opened DSA proceedings on 26 Jan 2026; Ireland's DPC opened a GDPR inquiry on 17 Feb 2026; French police raided X's Paris office (reported Feb 2026). Apple privately threatened in January 2026 to remove Grok from the App Store, then approved a revised version (reported 15 Apr 2026); NBC News later found similar images could still be made.
Age limit
13+ with parent or guardian permission for 13-17 per xAI terms (via search result, terms page blocked); not enforced beyond a self-reported birthdate (Common Sense Media). UK App Store rates it 16+, Google Play PEGI 18

In court

Lawsuits and regulator actions involving xAI. Claims are the claimant's allegations unless a court or regulator has ruled. The privacy score above was reviewed against this record. Who runs xAI: Elon Musk.

  • The plaintiffs allege Grok generated non-consensual nude and sexual images of them.

    N.D. Cal. · 5:26-cv-00772 · 28 Sep 2026: court denied motions to stay and transfer.

  • Skaggs alleges Grok disclosed private queries to third parties without users' consent.

    N.D. Cal. · 5:26-cv-04550 · 13 Aug 2026: court denied xAI's transfer motion.

  • Doe 1 alleges xAI used child sexual-abuse material in training and that Grok generated further abusive images.

    N.D. Cal. · 5:26-cv-09016 · 9 Oct 2026: court allowed a joint sur-reply filing.

  • Baltimore alleges deceptive trade-practices violations over Grok's image-editing capabilities.

    D. Md. · 1:26-cv-02103 · 27 Aug 2026: xAI filed a corporate disclosure statement; dismissal motion remains unresolved.

  • The ICO is investigating whether xAI and X complied with data-protection law in Grok's development and deployment.

    UK Information Commissioner's Office · No public reference number · 3 Feb 2026: ICO opened formal investigations; no outcome announced.

  • Ofcom is investigating whether X met UK Online Safety Act duties after reports involving Grok-generated imagery.

    Ofcom · No public reference number · 15 Jan 2026: Ofcom said its formal investigation remained ongoing.

  • The Commission is investigating whether X assessed and mitigated EU risks from deploying Grok functions.

    European Commission · DSA proceedings against X · 26 Jan 2026: Commission opened proceedings; outcome not prejudged.

  • The DPC is examining whether X complied with GDPR obligations concerning Grok-generated sexualised images involving EU/EEA data subjects.

    Irish Data Protection Commission · No public reference number · 17 Feb 2026: DPC announced inquiry; no decision published.

FAQ

Is Grok free?

Yes, there is a free plan with usage limits. Paid SuperGrok plans in the UK App Store start at £10.00 (SuperGrok Lite) and the standard SuperGrok is £30.00.

Is Grok safe for kids?

No. Common Sense Media rates Grok 'Unacceptable' and recommends no use by anyone under 18, saying age checks rely on a self-reported birthdate and Kids Mode does not reliably filter content.

Why is Grok being investigated in the UK?

In January 2026 Grok was used to create sexualised images of real people, including children. Ofcom opened an Online Safety Act investigation into X on 12 January 2026 and the ICO opened data protection investigations into X and xAI in February 2026; neither had concluded when checked.

Does Grok use my chats for training?

Secondary sources say yes by default. It can be turned off under Settings, Data Controls, 'Improve the model', and Private Chat mode is not used for training. xAI's own privacy policy could not be read to confirm this.

Grok vs ChatGPT?

Tom's Guide narrowly preferred Grok 4.1 over ChatGPT-5.2 in December 2025, mostly for detail. ChatGPT has fewer safety and regulatory problems and a cheaper standard paid plan in the UK.

Scores are based on published evidence, not lab tests. How we score.