StartStep AI United Kingdom Online Safety Compliance Statement
Version: V1.6 Effective: This version takes effect on the date it is published in the Rules Centre.
Last Updated: 10 September 2026
Effective Date: This version takes effect on the date it is published in the Rules Centre.
Provider: Noviq Pte. Ltd.
UEN: 202607454K
Registered Office: 68 Circular Road, #02-01, Singapore 049422
Official Domain: startstep.ai
Content and Safety Reports: support@startstep.ai
Emergency Notice: StartStep AI is not an emergency service. Contact 999 or the appropriate local emergency service where there is an immediate danger to life or safety.
Introduction
This United Kingdom Online Safety Compliance Statement (the “UK Online Safety Statement”) explains how Noviq Pte. Ltd. (“Noviq,” “StartStep AI,” “we,” “us” or “our”) approaches duties that may apply under the United Kingdom Online Safety Act 2023 (the “OSA”) to regulated parts of StartStep AI made available to people in the United Kingdom.
It supplements the StartStep AI Content and Community Rules, Complaints, Reports and Appeals Rules, AI Features and Responsible Use Notice, Global Privacy Notice and United Kingdom Privacy Supplement. Those documents remain applicable. This Statement does not create a broader monitoring, removal, response-time, age-verification or compensation commitment than the OSA or the StartStep AI Rules require.
StartStep AI accounts and authenticated Services are only for persons aged 18 or older. Parental or guardian consent does not lower that minimum age. This contractual rule does not replace a children's access assessment or any child-safety measure legally required for an in-scope service.
1. Scope and Service Classification
1.1 Regulated parts of StartStep AI
StartStep AI may enable Users to generate, upload or share content that can be encountered by other Users. Depending on the production design and legal tests, regulated user-to-user functionality may include:
- public posts, comments, replies, reviews and community discussions;
- published Projects, Project descriptions and contribution material;
- questions, answers and support material shared between Users or through a Project Host;
- profiles, ratings, reactions and other User-generated fields visible to others; and
- files, links, media or AI-assisted content that a User chooses to publish or share with another User.
Only functions actually enabled in production are within this description. Drafts, private prompts and Outputs visible only to the requesting User do not become user-to-user content merely because an AI system generated or processed them. If a User publishes, sends or shares that material through an in-scope function, the relevant user-to-user rules may then apply.
1.2 Search and AI functions
Internal navigation, retrieval or Project discovery may form part of another regulated function. It is not described as a separate regulated public search engine unless the applicable legal test is met. A private AI interaction is not treated as a regulated user-to-user service solely because an AI provider processes the request.
If UK law is amended or a feature changes so that a private AI, search, Agent or other function becomes separately regulated, Noviq may update the classification, assessment and measures for that function before or within the legally required period.
1.3 Function-by-function application
The OSA applies according to the actual service, functionality, UK link and statutory exemptions. A duty applying to one regulated part does not automatically extend every obligation to an unrelated or exempt part.
Noviq may add, restrict, redesign, geoblock, suspend or discontinue a function where reasonably necessary to manage legal, safety, technical or operational risk. Nothing in this Statement concedes a category designation or statutory status that has not been determined under the OSA.
1.4 Project discussion visibility
At initial release, Project discussions are visible to Users with the corresponding unlocked Project entitlement and authorised service personnel. General User-to-User messaging and Project ratings are not offered. This limited audience does not by itself exclude user-to-user functionality from the Online Safety Act. Service scope, exemptions and applicable duties are assessed against the actual design and UK links.
2. Our Online Safety Approach
2.1 Proportionate systems and processes
For an in-scope user-to-user service, we use or take proportionate measures designed to:
- prevent Users from encountering priority illegal content, as required;
- mitigate and manage the risk that the Service is used to commit or facilitate priority offences;
- mitigate and manage risks identified in the applicable illegal-content risk assessment;
- minimise the time for which priority illegal content is present;
- swiftly take down illegal content when Noviq becomes aware of it and the applicable legal test is met; and
- provide accessible reporting and complaints procedures.
Measures are selected according to the Service's type, size, functionality, user base, risk assessment and technical feasibility. Not every measure applies to every feature or at every stage of growth.
2.2 Codes and alternative measures
Noviq may implement applicable measures recommended in Ofcom's Codes of Practice. We may instead use an alternative measure where permitted, provided we make and retain the assessment and records required to show how it fulfils the relevant duty.
Compliance with a Code does not prevent us from using additional controls. A recommendation aimed at a large, high-risk or differently designed service is not automatically adopted by StartStep AI where it is not applicable or proportionate.
2.3 Internal accountability
Noviq assigns an individual internally accountable to senior management for relevant online-safety compliance. We need not publish that person's name unless law or a competent authority requires it. Operational work may be carried out by authorised employees, contractors, providers and AI-assisted systems, but Noviq retains responsibility for the duties applicable to it.
3. Illegal Content and Prohibited Conduct
3.1 Content within the illegal-harms assessment
Our internal assessment covers each kind of priority illegal content specified by the OSA and relevant non-priority illegal content reasonably indicated by the Service. Depending on context, this includes risks relating to terrorism; child sexual exploitation and abuse; threats, harassment, stalking and hate offences; controlling or coercive behaviour; sexual exploitation; intimate-image abuse and cyberflashing; fraud and financial crime; illegal drugs, weapons or regulated goods; human trafficking or unlawful immigration assistance; and encouragement or assistance of serious self-harm.
This summary is not an exhaustive statement of UK criminal law. Content may be illegal under another offence or prohibited by the StartStep AI Rules even where it does not fall within a priority category.
3.2 Community prohibitions
Users must comply with the Content and Community Rules. Noviq may prohibit or restrict content and conduct that is unlawful, facilitates an offence, creates material safety or fraud risk, infringes rights, undermines Service integrity or otherwise violates those Rules.
The OSA does not require StartStep AI to host all lawful content. Subject to mandatory law, we may restrict content that is lawful but violates a clearly applicable StartStep AI rule. We may also retain, label, reduce visibility, disable interaction, age-gate or impose another proportionate measure instead of removal.
3.3 No approval by non-detection
Content is not approved merely because it was published, generated, indexed, recommended or not detected. We may reconsider it after a report, new evidence, a provider or authority signal, legal change, risk-assessment update or later review.
4. Risk Assessments and Change Control
4.1 Illegal-content risk assessment
For each in-scope regulated service, Noviq carries out and records a suitable and sufficient illegal-content risk assessment within the period required by the OSA. The assessment considers how Users may encounter illegal content and, for user-to-user functions, how the Service may be used to commit or facilitate priority offences.
The assessment is based on reasonably available evidence, Ofcom's current risk profiles and register of risks, the Service's functionalities, user base, business model, recommendation and sharing design, reports, incidents and existing controls.
4.2 Review triggers
We keep an assessment up to date and review it when required, including before a significant change to the design or operation of the relevant Service. A significant change may include a new public-sharing function, direct messaging, live streaming, anonymous access, material recommendation change, material expansion of AI publication, a new territory or a material change in user scale or risk.
We may apply a temporary restriction while an assessment, investigation or mitigation is completed. Product launch approval may be withheld where the responsible owner has not completed a required assessment.
4.3 Confidential records and categorised services
The full risk assessment, evidence, internal thresholds, security controls and operational playbooks are confidential compliance records unless disclosure is legally required. This public Statement is not represented as the statutory risk-assessment summary of a categorised service unless Noviq expressly identifies it as such after a relevant categorisation.
If Ofcom places a StartStep AI service on a statutory register or gives it Category 1, 2A, 2B or another designation, Noviq will apply the additional assessment, submission, transparency, user-empowerment or other duties that legally attach to that designation. No such status is assumed merely because this Statement is published.
5. Safety Measures and Moderation
5.1 Measures we may use
Subject to the applicable risk assessment and proportionality, measures may include:
- publication controls, rate limits and limits on links, files or automated posting;
- keyword, hash, classifier, model or behavioural signals;
- human review, sampling, escalation and specialist referral;
- reduced distribution, warning labels, interaction limits and recommendation controls;
- removal, access restriction, feature restriction, account suspension or termination;
- blocking, muting, comment controls or other User tools where available or required;
- fraud, identity, device, payment and account-integrity controls;
- preservation of evidence and reports to authorities or designated bodies; and
- guidance and support information appropriate to the identified risk.
We are not required to use every listed measure on every item. We may change a measure as risk, evidence, law, technology and Service design evolve.
5.2 Automated systems and human review
Automated systems may prioritise, block, label, restrict or refer content, accounts or conduct. They may make mistakes and are not a guarantee that all harmful content will be detected.
We provide human review or another safeguard where required by law or the applicable complaints process. An urgent protective restriction may remain in place during review where removal would create a credible risk of illegality, exploitation, physical harm, fraud, retaliation, evidence loss or control evasion.
5.3 No general pre-approval promise
Unless law specifically requires a measure, Noviq does not promise to monitor or approve every item before publication, read every private communication, preserve all content indefinitely or prevent every instance of unlawful or harmful conduct. We may use reactive and proactive measures in the combination reasonably selected for the Service.
5.4 Consistent application of published rules
We apply the safety and moderation provisions in the Rules consistently as required by applicable law. Consistency does not require identical outcomes where context, evidence, severity, reach, intent, recurrence, User role, technical feasibility or risk differs.
6. Reporting Illegal or Harmful Content
6.1 Who may report
A User or affected person in the United Kingdom may report suspected illegal content or a matter within an applicable OSA reporting duty. A person does not need to create an account merely to report content where the law requires access for non-Users or affected persons.
An affected person may include a person who is the subject of content, belongs to a targeted class or group, has responsibility for an affected child, or assists an adult who needs help using the Service, where the statutory definition applies.
6.2 How to report
Use the report control attached to the content or account where available, or:
Content and Safety Report: support@startstep.ai
A useful report identifies the content or account, URL or content identifier, date, reason, relevant UK connection and supporting context. We may request additional information reasonably necessary to locate and assess the matter.
Do not upload, forward or resend suspected child sexual abuse material or non-consensual intimate imagery in a report. Provide the location or identifier and follow any secure instructions. Contact emergency services where there is an immediate danger.
6.3 Triage and action
We may use automated triage and prioritise credible reports involving child safety, imminent physical harm, terrorism, intimate-image abuse, serious fraud or other high-severity risk. Priority does not mean every allegation is accepted as true.
We may review the reported content, related context, account history, technical records, prior decisions and reasonably available evidence. Depending on the result, we may take no action, request information, label, restrict, remove, preserve, suspend, refer, report or take another lawful measure.
7. Complaints and Appeals
7.1 Eligible complaints
The StartStep AI Complaints, Reports and Appeals Rules provide the general procedure. Where required by the OSA, the procedure is available for eligible complaints concerning:
- how StartStep AI handles a report of illegal content;
- a decision to take down, restrict or leave up content;
- an action affecting a User's access to, or use of, the Service because of User-generated content;
- application of a relevant safety or terms-of-service provision;
- failure to implement a legally required reporting, complaints or safety measure; and
- child-safety content or measures where the protection-of-children duties apply.
7.2 Submission and review
Submit a complaint or appeal through:
Identify the decision, report or content and explain the claimed error, missing context, applicable right or requested outcome. We may verify account ownership or authority and may combine duplicates.
We will operate the procedure in accordance with applicable law. No fixed public response period is promised unless law, a specific notice or another Rule requires one. We may prioritise by risk and complexity and keep an urgent measure in place during review.
7.3 Outcome and reasons
We may uphold, vary or reverse a decision, restore or remove content, change an account measure, request corrective action, continue investigation or refer the matter externally. We provide notice, reasons and further review information to the extent required by law.
We may withhold or limit information where disclosure would be unlawful, identify or endanger a reporter or victim, reveal another person's data, prejudice an investigation, disclose privileged or confidential material, expose security methods or facilitate evasion.
7.4 No final legal determination
A moderation or complaint decision is not a court judgment and does not finally determine criminality, civil liability or rights ownership. A person may use a competent court, authority or statutory route. Ofcom does not ordinarily decide individual takedown or reinstatement complaints merely because it regulates the Service.
8. Child Access and the 18+ Rule
8.1 Children's access assessment
For every in-scope user-to-user or search service, Noviq completes and records the children's access assessment required by the OSA. A contractual statement that Users must be 18, a date-of-birth field or an ordinary self-declaration does not by itself establish that children cannot normally access the Service.
We may conclude that children are not likely to access a Service only on the evidence and legal standard applicable at the time. Where the conclusion depends on preventing child access, the relevant age-assurance method must meet the legally required effectiveness standard.
8.2 Reassessment
If an assessment concludes that a Service is not likely to be accessed by children, we repeat it within 12 months or earlier when required, including after a significant change, evidence of child use or a relevant change in Ofcom guidance.
If all or part of the Service is likely to be accessed by children, we complete the required children's risk assessment and implement proportionate protection-of-children measures within the required period. We may restrict the relevant feature, deploy age assurance or decline to launch it until those measures are ready.
8.3 Age assurance and privacy
Where age assurance is required or reasonably selected, we may use age verification, age estimation, account or payment signals or another lawful method appropriate to the risk. A method required to be highly effective must be technically accurate, robust, reliable and fair under the applicable guidance.
We will provide the privacy notice and choice required for the method, minimise data, assess the provider and offer an appropriate alternative where legally required or reasonably practicable. The use of age assurance does not authorise unrelated identity profiling.
8.4 Underage accounts
If we reasonably believe that an account belongs to a person under 18, we may restrict or close it, disable publication and request proportionate age evidence. Adults must not help a child evade an age check or give a child access to an adult account.
9. Child Sexual Exploitation and Abuse Reporting
9.1 Zero tolerance
Child sexual abuse material, grooming, sexual exploitation, sextortion, trafficking, instructions to obtain abuse material and attempts to contact or endanger a child are prohibited. This applies to real and, to the extent prohibited by law or the Rules, AI-generated, synthetic, illustrated or altered material.
9.2 Required reports
Where the OSA reporting duty applies, Noviq operates systems and processes designed, so far as possible, to report detected and unreported child sexual exploitation and abuse (“CSEA”) content through the legally applicable route and within the required format and timeframe.
As a non-UK provider, Noviq will assess the applicable UK link and whether the content has already been reported through an eligible foreign reporting regime. We will not knowingly submit a duplicate report to the NCA and an equivalent foreign body where the law instructs providers not to duplicate reporting.
The duty concerns CSEA content StartStep AI detects; it does not mean this public Statement promises that StartStep AI will detect every item. Detection may result from a User report, provider or child-protection signal, hash matching, automated control or human review.
9.3 Secure escalation and preservation
Authorised personnel may preserve and securely transmit content, metadata and User information required for a report, safeguarding action, legal hold or investigation. Access is limited according to role and legal need. Users and ordinary support personnel must not download or circulate suspected abuse material.
Where required, Noviq will register with and use the National Crime Agency's CSEA Industry Reporting Portal or maintain the permitted alternative reporting route. A current or imminent danger should also be reported to the appropriate emergency authority.
10. Privacy, Freedom of Expression and User Rights
10.1 Privacy and safety processing
We may process Community Content, prompts or Outputs shared with others, metadata, account and device information, risk signals, reports, complaint material, moderation decisions and evidence for safety, moderation, CSEA reporting, legal compliance, claims and Service integrity. The Global Privacy Notice and UK Privacy Supplement explain the lawful bases, recipients, transfers, retention and rights.
Safety processing is limited and protected in proportion to risk. A privacy request does not require deletion of evidence that Noviq is legally required or permitted to preserve, but data is not retained indefinitely merely by labelling it “safety data.”
10.2 Freedom of expression and privacy considerations
When designing and applying proportionate safety systems, we consider the importance of protecting Users' freedom of expression and privacy as required by the OSA. These considerations do not create an unconditional right to publish, receive, recommend or retain content through a private Service.
We may take an interim measure on credible evidence without first proving an offence to a criminal standard. We may consider context, public interest, journalistic or educational purpose, satire, counterspeech, documentary evidence and the risk of misuse where relevant.
10.3 Confidentiality of controls and reports
We do not disclose confidential classifiers, keywords, hashes, model prompts, review thresholds, fraud signals, reporter identities, security architecture or evasion-sensitive guidance except where validly required. Transparency obligations do not require disclosure that would undermine safety, security, privacy, investigations or third-party rights.
11. Authorities, Ofcom and Legal Requests
Noviq may comply with a valid information notice, access restriction, preservation direction, court order, law-enforcement request or other binding requirement. We may challenge, narrow or seek clarification of a request where lawful and appropriate, but do not promise to challenge every request.
We maintain records and provide information to Ofcom where required. Publishing this Statement does not mean Ofcom has approved StartStep AI, its risk assessment or its measures.
Users may obtain information about Ofcom at ofcom.org.uk. Ofcom may use reports and intelligence for regulatory purposes but is not a substitute for StartStep AI's individual content-reporting or appeal process, a court or emergency services.
12. Changes and Service Measures
We may update this Statement or the related safety measures for legal, regulatory, risk, security, product, provider, technical, operational or organisational reasons. The updated version will state its publication and effective dates.
We may act immediately where delay would materially increase illegality, child-safety risk, physical harm, fraud, evidence loss, security compromise or regulatory breach. Where a change materially affects Users and notice is legally required, we will provide the required notice.
13. Contact
Provider: Noviq Pte. Ltd.
UEN: 202607454K
Registered Office: 68 Circular Road, #02-01, Singapore 049422
Content and Safety Reports: support@startstep.ai
Legal Notices: legal@startstep.ai
Use the in-product report control for a specific item where available. A public post, app-store review, social-media comment or message to a Project Host is not formal notice to Noviq.