AI at PubNub

AI Responsible Use and Security at PubNub

Our Approach

At PubNub, we use artificial intelligence (AI) and machine learning responsibly to enhance the reliability, performance, and security of our real-time data-streaming platform. Our goal is to leverage these technologies in a transparent, compliant, and secure way that prioritizes customer trust, privacy, and control.

Where We Use AI

We apply AI and machine learning techniques across specific areas of our platform and operations to improve performance and user experience. Examples include:

check

Real-time anomaly detection for streaming data, monitoring patterns, and alerting on irregular behavior.

check

Smart routing and optimisation of data flows to deliver lower latency and higher reliability.

check

Automated operational support: using machine-learning models to surface incidents, predict capacity usage, and assist with system maintenance.

check

Improved developer experience: offering intelligent SDKs, code assistance, and usage analytics that leverage AI-driven insights.

Governance & Controls

To ensure safe and effective use of AI, we have established the following governance framework:

check

An AI Use Policy that applies across internal teams (engineering, product, operations) and third-party contractors, specifying permitted and prohibited uses of AI tools and models.

check

An Onboarding Process for any new AI tool or use case: prospective uses are evaluated via a questionnaire and reviewed by a multidisciplinary team, including Compliance, Engineering, Legal, and Security.

check

A central Repository of Approved AI Tools and Use-Cases: once onboarded, the tool/use-case is catalogued for transparency, reuse, audit readiness, and continuous oversight.

check

Ongoing quality-control and monitoring: we require human-in-the-loop reviews where appropriate, check for bias or inaccuracies in model outputs, and ensure alignment with our values and obligations.

Data & Privacy Considerations

When AI features process personal data or metadata, PubNub applies the same rigorous privacy protections used across our platform. Data is encrypted in transit and at rest, processed in accordance with data minimization principles, and governed by strict access controls.

PubNub does not use customer data to train or fine-tune any AI models. Data processed by AI features such as Analyze with AI is used only for the requested analysis and is never stored, retained, or shared with AI providers.

All AI-related use cases involving sensitive or regulated data undergo a privacy and security risk assessment before deployment to ensure appropriate safeguards are in place.

Security & Compliance

All AI-enabled systems follow the same security standards and controls that govern the broader PubNub platform, including secure development lifecycle, vulnerability management, incident response, and audit logging.

PubNub continuously monitors the AI ecosystem to identify and mitigate risks such as bias, model drift, or adversarial inputs. Our framework aligns with recognized standards, including the NIST AI Risk Management Framework and ISO/IEC 42001, ensuring our AI use remains safe, transparent, and compliant.

Innovation & Transparency

PubNub believes that AI can enhance real-time communication and data experiences when used responsibly. We foster innovation through controlled experimentation in a regulated environment, ensuring full transparency regarding our use of AI.

PubNub is excited to stay ahead of evolving AI regulations, such as the EU AI Act and upcoming U.S. frameworks, ensuring our practices are not only adaptive but also future-ready and responsible.

expand_less

Does PubNub use AI in its products and services?

Yes. PubNub uses AI responsibly to enhance data understanding, developer productivity, and user experiences. AI features are optional and are only activated when a customer enables them. This ensures that every organization maintains full control over its data, workflows, and AI use across PubNub’s platform.

Yes. PubNub uses AI responsibly to enhance data understanding, developer productivity, and user experiences. AI features are optional and are only activated when a customer enables them. This ensures that every organization maintains full control over its data, workflows, and AI use across PubNub’s platform.

expand_more

When and how do customers activate AI within PubNub?

AI is available as an opt-in feature through PubNub’s Insights and Illuminate products. Customers decide when to enable capabilities such as Analyze with AI, which provides data summaries, anomaly detection, and recommendations. These features remain off by default until the customer activates them, ensuring AI use always aligns with organizational policies and compliance requirements.

expand_more

Is AI a core or supplementary element of PubNub?

AI is a supplementary enhancement to PubNub’s core real-time Data Stream Network. The PubNub infrastructure operates fully without AI, which is designed to add contextual intelligence, summaries, and insights for customers who choose to enable it.

expand_more

Does PubNub use Generative AI, and how is it applied?

Yes. Generative AI capabilities in Insights and Illuminate are used to summarize, classify, and analyze data. PubNub also supports AI-powered code generation, helping developers quickly build PubNub-based applications. Generative AI is always used under strict data protection controls and never for autonomous decision-making.

expand_more

Does PubNub build or train its own large language models (LLMs)?

No. PubNub does not build or train proprietary models. Instead, we leverage pre-trained, enterprise-grade models from trusted providers such as OpenAI, Anthropic, and Hugging Face. Each model is carefully evaluated for performance, data security, and compliance with our Responsible AI standards.

expand_more

How does PubNub ensure customer data is not used for AI model training?

Customer data is never used to train or fine-tune AI models. All PubNub AI integrations process data in real time only for the specific, authorized purpose defined by the customer. Data is encrypted using AES-256, anonymized where possible, and deleted after use. PubNub enforces contractual and technical safeguards to prevent third parties from storing or reusing customer data.

expand_more

Who oversees AI governance and compliance at PubNub?

A cross-functional AI Governance Committee oversees PubNub’s AI. It includes representatives from Compliance, Legal, Security, Product, Finance, and Engineering. This committee ensures that all AI features adhere to internal standards, international frameworks such as NIST AI RMF and ISO/IEC 42001, and evolving global regulations.

expand_more

What Responsible and Ethical AI policies guide PubNub’s approach?

PubNub’s Responsible AI framework is founded on the principles of transparency, accountability, privacy, and fairness. Every AI capability undergoes review and validation to ensure ethical deployment. PubNub communicates clearly when AI is in use and provides explainable outputs to maintain user trust and understanding.

expand_more

Does PubNub use AI-generated results or decisions without human oversight?

No. All AI outputs are designed to assist human users, not replace them. Customers review and approve all insights and recommendations generated by AI. PubNub does not implement fully automated decision-making systems.

expand_more

How can customers verify when AI is being used?

PubNub provides full transparency. When features such as Analyze with AI are activated within Insights or Illuminate, the interface clearly indicates AI-assisted insights. Documentation and in-product labeling make it clear when and how AI contributes to an analysis, ensuring users are always aware of its role.

The use of AI is also governed by each customer’s contractual agreement with PubNub, which defines how data is processed, what features are enabled, and the safeguards in place. This agreement ensures that customers maintain complete visibility and control over AI-related functionality within their environments.

expand_more

Does PubNub use third-party AI providers, and what safeguards are in place?

Yes. PubNub integrates with OpenAI, Anthropic, and Hugging Face under contractual agreements that prohibit the storage or reuse of customer data. Each integration is reviewed by PubNub’s Vendor Risk Management process and Security teams to ensure alignment with our privacy, compliance, and security standards.

expand_more

How does PubNub ensure transparency, reliability, and fairness in AI systems?

PubNub continually monitors the performance and reliability of AI integrations to prevent bias, drift, or degradation. Explainability and traceability are built into each system, allowing customers to understand the reasoning behind AI-generated insights. PubNub’s governance process ensures that fairness, accuracy, and ethical standards remain central to every AI feature released.