Provably Fair vs RNG Certified: What UKGC, MGA and GSC Actually Accept
Neither the UKGC, MGA nor GSC accepts provably fair in place of RNG certification. See what each regulator requires and how to layer both without audit gaps.
The UKGC, the MGA and the Isle of Man Gambling Supervision Commission (GSC) do not accept a provably fair cryptographic scheme as a substitute for independent RNG and game testing. As of November 2026, the UKGC’s Remote Gambling and Software Technical Standards (RTS 7), MGA Directives 2 and 3 of 2018 and the GSC’s approved test facility requirement all place the assurance burden on accredited laboratory review or Authority approval of the RNG, its statistical output and return to player. A provably fair verifier can sit on top of a certified game as a transparency feature, and nothing in the texts reviewed forbids that, but it cannot discharge a licence condition.
This comparison rests on a read of the RTS (published 2 February 2021, last updated 31 October 2025), MGA Directive 2 (V3, January 2023), Directive 3 (V2, October 2021), GLI-19 v3.0 and the published description of the GSC testing regime. Licensees should take advice from qualified legal counsel in each jurisdiction before relying on it for a product or licensing decision. For detailed regulatory requirements specific to your licence type, refer to the Key Resources section at the end of this article.
Can a licensee rely on provably fair instead of a certified RNG?
No. None of the three regulators recognises a provably fair scheme as an alternative route to compliance. Each requires the RNG and game logic to be assessed by an independent test laboratory or, in Malta, approved by the Authority before release, and a provably fair layer is permitted only as an addition to that assessed baseline.
A provably fair scheme in the usual commit-reveal form works as follows. The operator publishes a hash of a secret server seed before play, the player supplies or accepts a client seed, and each bet derives its outcome from the two seeds plus a nonce. After the seed is rotated, the operator reveals the server seed so the player can recompute the hash and the outcomes. A certified RNG is a different assurance object. An independent laboratory reviews the source code of the randomness, scaling and mapping algorithms, runs statistical tests on the final outcome distribution, and issues a certificate the regulator relies on.
The RTS, Directives 2 and 3 and GLI-19 v3.0, in the versions reviewed, contain no provision that names provably fair or treats on-chain or hash-based verification as equivalent to laboratory testing. A licensee proposing a provably fair game as its primary fairness control is therefore asking each regulator for a deviation.
SBC News reported on 11 September 2026 that the SBC Summit Web3 and Crypto Academy would include a session titled “Provably Fair: Can Players Really Verify Every Outcome?”, comparing RNG models with on-chain verification and examining the limitations of the latter. The session is trade programming, not regulatory guidance.
UKGC: RTS 7 and the test house framework
Licence condition 2.3.1 of the Licence Conditions and Codes of Practice (LCCP) makes the RTS binding on remote operating licence holders and gambling software licence holders. RTS 7 applies to gaming (including bingo), lotteries and betting on virtual events, and its stated aim is to ensure that games and other virtual events operate fairly.
“Random number generation and game results must be ‘acceptably random’. Acceptably random here means that it is possible to demonstrate to a high degree of confidence that the output of the RNG, game, lottery and virtual event outcomes are random through, for example, statistical analysis using generally accepted tests and methods of analysis. Adaptive behaviour (that is, a compensated game) is not permitted.” (UKGC, RTS requirement 7A)
Source: UKGC, Remote gambling and software technical standards (RTS), RTS 7 Generation of random outcomes, requirement 7A, published 2 February 2021, last updated 31 October 2025.
The test in 7A is statistical and population-level. The implementation guidance lists the qualities an RNG should be able to demonstrate. Output should be uniformly distributed over the full range, and game outcomes should follow the expected or theoretical probabilities. The output should be unpredictable, which for a software RNG means it is computationally infeasible to predict the next number without complete knowledge of the algorithm and seed value. The stream must not cycle or synchronise with another instance, seeding and re-seeding must not introduce predictability, and any scaling must preserve these qualities.
Two adjacent requirements matter for hash-derived outcomes. The guidance to RTS 7B states that random numbers received by a game must be used in the order received and may not be discarded because of adaptive behaviour, and may be discarded only if they fall outside the range the virtual event requires, in which case an error must be logged and investigated. RTS 7C bars game designs that mislead customers about the likelihood of particular results, including near-miss substitution and simulations of real devices that do not reproduce real probabilities.
The RTS uses the language of external verifiability only for lotteries that determine a result from events outside the lottery. There, the outcome must be unpredictable and externally verifiable, and the guidance gives results published in local or national press as an acceptable example. That requirement does not extend to casino RNG outcomes, so it is not UKGC acceptance of player-side verification for slots or table games.
On testing, RTS paragraph 1.6 requires licensees responsible for procuring games testing to submit a Games testing annual audit, with detail in the UKGC testing strategy. The Commission’s test house framework consultation (response published 3 February 2020) stated that the Commission needs to be assured of test house independence, competence and suitability so that games are suitable for release to market. The separate RTS security chapter is based on Annex A of ISO/IEC 27001:2022, which replaced the 2013 edition, our analysis of ISO/IEC 27001 implementation errors in iGaming covers how those controls are typically misconfigured. The UKGC guide page for the RTS notes updates effective 30 October 2026.
Crypto payment rails raise a separate LCCP point. Condition 5.1.1 requires licensees to implement policies and procedures on customers’ use of cash and cash equivalents, naming digital currencies among the examples, designed to minimise money laundering risk. A provably fair crypto casino serving Great Britain without a UKGC licence sits outside the framework entirely and is an unlicensed supply matter.
MGA: game approval under Directives 2 and 3
The MGA regulates through the Gaming Act (Cap. 583), the Gaming Authorisations Regulations and the 2018 Directives, and it gates RNG-based games at the point of introduction rather than through a single technical standard. Directive 3 of 2018 (Gaming Authorisations and Compliance Directive, V2 October 2021) defines a game provider as a B2B licensee providing a critical gaming supply under point 3 of the First Schedule to the Gaming Authorisations Regulations.
Under article 19 of Directive 3, a B2C licensee must notify the Authority no later than five days after adding games, and that notification requirement does not apply where the games come from an authorised game provider with which it has a pre-existing relationship. Where a game uses a random number generator, or is based on a game engine, that the Authority has not already approved, prior written approval is required. Article 23(1) provides that an application for approval of a new game using an RNG or unapproved game engine must be accompanied by such documents and certification as the Authority may require, and article 23(2) reserves the Authority’s right to require an audit.
Directive 3 also defines “essential components” to include components hosting random number generators. Article 18(1) requires licensees to subject essential components to additional safeguards, and article 17(4) requires a live or real-time mirror server for essential regulatory data, accessible to the Authority. Our article on MGA system audit requirements maps how those obligations appear in a system audit evidence pack.
The Player Protection Directive (Directive 2 of 2018, V3 January 2023) adds outcome controls. Article 22(3) requires B2C licensees to ensure their systems monitor the average percentage returned to players as prizes for these random-selection games, and a licensee whose games are hosted and managed by an approved B2B licensee may have that obligation fulfilled by the B2B licensee. Article 23 allows the Authority to require a licensee or applicant to submit a certificate issued by a testing lab certifying compliance with the Directive. Article 24 empowers the Authority to audit, to request a report of average percentage returned over a specified period or number of plays, and to conduct any test required. Article 25 empowers it to require withdrawal of any game that does not comply.
The MGA’s Technical Infrastructure guidance for remote gaming (December 2015, v1) is principles-based and identifies accepted industry standards that the MGA considers to meet its requirements, while stating that deviations will be considered on the basis of risk and mitigation. That language is the nearest thing in the Maltese materials to an opening for novel verification mechanisms, and it operates through case-by-case submission to the Authority.
Isle of Man GSC: approved test facility model
Primary-source coverage is thinnest for the GSC, so the position below is stated narrowly. Published descriptions of the Isle of Man regime state that the Commission approves independent testing facilities to ensure the randomness of results of all games of chance, and that companies seeking an Isle of Man licence must test their software systems with one of the approved companies. GLI’s own company record states that it was accredited as an approved online gaming testing facility by the Gambling Supervision Commission.
The model is closer to a conventional approved-laboratory regime than to the MGA’s game-approval gate. Assurance of randomness is delegated to a facility the Commission has itself approved, so a provably fair mechanism has no standalone status. Whether the GSC would consider a provably fair design within an approved facility’s scope is not addressed in the materials reviewed. Licensees and applicants must read the current Isle of Man online gambling legislation and GSC guidance directly and raise any hash-based design with the Commission before build, with Manx counsel involved.
Provably fair versus certified RNG, mechanism by mechanism
The table below compares what each approach produces as evidence, as of November 2026. The right-hand column cites the provision that governs the certified approach. The comparison itself is analysis, not regulatory text.
| Dimension | Provably fair scheme | Laboratory-certified RNG | Governing reference |
|---|---|---|---|
| Evidence produced | Per-bet proof that the outcome follows from a committed seed | Test report on RNG, scaling, mapping and game logic | RTS 7A, GLI-19 v3.0 chapter 3 |
| Who verifies | Player or a third-party verifier tool | Independent test laboratory accepted by the regulator | GLI-19 3.2.1, GSC approved test facilities |
| Timing | After the bet, once the server seed is revealed | Before release, and again on material change | MGA Directive 3 art. 19 and 23 |
| Distribution and bias | Not tested by hash recomputation | Statistical tests on the final outcome after scaling, at 99% confidence | GLI-19 3.2.2, RTS 7A guidance |
| Source code review | Optional and often self-published | Required for core randomness, scaling and shuffling algorithms | GLI-19 3.2.1 |
| Return to player | Not demonstrated by seed verification | Verified in testing and monitored in operation | MGA Directive 2 art. 22 |
| Regulator visibility | Player level only | Certificate and report available to the regulator | MGA Directive 2 art. 23 |
Hash verification proves one narrow thing, that the revealed seed matches the earlier commitment and that the displayed outcome follows from the published derivation. It does not show that the derivation maps uniformly onto game outcomes, that the paytable delivers the stated return to player, or that the server seed was generated with adequate entropy. GLI-19 v3.0 requires the independent laboratory to review source code for bias, implementation errors and malicious code, and to evaluate the final outcome output with statistical tests such as chi-square, runs and serial correlation tests at a 99% confidence level. It also requires that each possible RNG selection be equally likely to be chosen, and where the game design specifies a non-uniform distribution, that the final outcome conform to the intended distribution.
Regulators already use cryptographic hashing for a different purpose. GLI-19 v3.0 section 2.3.2 requires the system to verify, at least once every twenty-four hours and on demand, that critical control program components are authentic copies of approved components, using a cryptographic hash algorithm that produces a message digest of at least 128 bits. That control authenticates software, and it does not prove any individual game outcome. GLI-19 also defines a cryptographic RNG as one resistant to attack by an intelligent attacker with modern computational resources who knows the source code or algorithm. Our comparison of GLI-19 and GLI-33 sets out where GLI-19 sits in a certification path.
What do UKGC, MGA and GSC accept as evidence of fairness?
As of November 2026, each regulator accepts laboratory-based evidence tied to its own instrument, and none expressly recognises provably fair verification. The table summarises the acceptance position and the route open to a licensee proposing a novel mechanism.
| Regulator | Governing instrument | Fairness evidence accepted | Provably fair expressly recognised | Route for a novel mechanism |
|---|---|---|---|---|
| UKGC | LCCP condition 2.3.1, RTS 7 (updated 31 October 2025) | Statistical demonstration of acceptable randomness, Games testing annual audit where applicable (RTS 1.6) | No provision in the text reviewed | Meet RTS 7 in full and agree testing scope under the testing strategy |
| MGA | Gaming Act (Cap. 583); Directive 3 (V2 October 2021) art. 19, 23, Directive 2 (V3 January 2023) art. 22 to 25 | Prior written approval of unapproved RNGs and game engines, certification as the Authority requires, testing lab certificate | No provision in the text reviewed | Application under Directive 3 art. 23, Technical Infrastructure guidance considers deviations on a risk basis |
| GSC | Isle of Man online gambling legislation, GSC approved test facilities | Software tested by a GSC-approved facility to ensure randomness | Not located in the materials reviewed | Direct engagement with the Commission before build |
The position is consistent with other regimes that name the same control. The AGCO’s ITL certification policy requires all games, random number generators and components that accept, process, determine the outcome of, display and log player bets to be certified by an AGCO-registered independent test laboratory. The AGLC’s Standards and Requirements for Internet Gaming (2026-03-17) require certification by an AGLC-registered testing facility before deployment in the Alberta market.
Decision frame: which route for which licensee profile
A UK-facing remote operating licensee or gambling software licensee should treat RTS 7 and a test house certificate as the licensing basis and add provably fair verification only after the certified game is stable. RTS 3 (Rules, game descriptions and the likelihood of winning) and RTS 7C apply to how a game and its fairness are described to customers, so a marketing claim of “100% provably fair” should be reviewed against both before launch.
An MGA B2C or B2B licensee should run any provably fair wrapper through the game approval path. If the wrapper changes how outcomes are derived, it may constitute a new RNG or game engine requiring prior written approval under Directive 3 article 19, this is analysis, and the position should be confirmed with the Authority. A B2C licensee using a hosted B2B game provider can have the article 22(3) return-to-player monitoring obligation fulfilled by that provider.
An Isle of Man applicant should ask its approved test facility whether the engagement scope covers seed generation, the commit-reveal derivation and the mapping of hash output to game outcomes, and should confirm the answer with the GSC in writing.
A crypto-native platform seeking any of the three licences should assemble a laboratory evidence pack covering the mapping from hash output to outcome, server seed generation and scaling, with the provably fair verifier presented as a supplementary feature. For UK licensees, LCCP condition 5.1.1 requires the digital currency payment policies to sit in the same file.
Key requirement: as of November 2026, an MGA licensee needs prior written approval for any game using an RNG or game engine the Authority has not approved (Directive 3, article 19), and a UKGC licensee must demonstrate acceptable randomness under RTS 7A. Neither obligation is satisfied by seed verification alone.
Frequently asked questions
Does the UKGC accept provably fair as proof of fairness?
No. RTS 7A requires random number generation and game results to be acceptably random, demonstrable by statistical analysis using generally accepted tests. The RTS text reviewed contains no provision recognising provably fair verification as an alternative.
Does the MGA require RNG certification?
Directive 3 of 2018 requires prior written approval for any game that uses an RNG or game engine the Authority has not already approved. The application must include such documents and certification as the Authority may require, and Directive 2 article 23 lets the Authority demand a testing lab certificate.
Can a licensed operator offer provably fair verification alongside certified games?
The texts reviewed do not prohibit it. The certified RNG remains the licensing basis, and any customer-facing fairness claim should be reviewed by counsel against the likelihood-of-winning and misleading-design requirements in RTS 3 and RTS 7C.
What does the Isle of Man GSC require for RNG testing?
Published descriptions state that the Commission approves independent testing facilities and that licence applicants must test their software systems with one of them. Applicants should confirm current wording and scope with the GSC directly.
Key Resources
UKGC, Remote gambling and software technical standards (RTS), including RTS 7, RTS 3 and paragraph 1.6: https://www.gamblingcommission.gov.uk/standards/remote-gambling-and-software-technical-standards
UKGC, Licence Conditions and Codes of Practice, condition 2.3.1 and condition 5.1.1, available from the Gambling Commission website.
MGA, Directive 2 of 2018 (Player Protection Directive, V3 January 2023), Directive 3 of 2018 (Gaming Authorisations and Compliance Directive, V2 October 2021) and Technical Infrastructure guidance for remote gaming (December 2015, v1): https://www.mga.org.mt/
GLI-19 Standards for Interactive Gaming Systems v3.0, chapter 3 and section 2.3.2: https://gaminglabs.com/gli-standards/
Matt Denney
Editorial · gamingcompliance.io
Reads the primary source so you don't have to. Fifteen years inside iGaming compliance: operator, supplier, and crown-corporation lottery.