How to scale AI responsibly

Artificial intelligence is moving faster than most organisations can govern it. Fewer than 1% of organisations have fully operationalised responsible AI in a comprehensive and anticipatory manner. (World Economic Forum)

AI governance is no longer an abstract ethics issue.

Organisations that build responsible AI into how they work will scale faster, earn trust more consistently and avoid preventable failures as AI becomes more customer-facing.

For marketing leaders, this shift is especially urgent.

Marketing is one of the most exposed functions in the enterprise: it shapes public messages, customer interactions, personalisation decisions and brand experiences at scale. (Stanford HAI)

The gap between policy and practice

Most organisations are no longer ignoring responsible AI. The more revealing problem: many still treat it as a policy statement rather than an organisational capability.

The 2026 AI Index reports that the share of businesses with no responsible AI policies fell from 24% in 2024 to 11% in 2025. Formal adoption is growing. (Stanford HAI)

But the same research shows that maturity is still at its early stage. On a four-point scale, the global average reached 2.3 in 2025. Most organisations are still putting responsible AI practices in place, not yet implementing them as comprehensive systems.

The barriers go wider than training. Knowledge and training gaps remain the most-cited obstacle, followed by budget constraints, regulatory uncertainty and technical limitations. For more autonomous agentic systems, security and risk concerns outweigh the rest. (McKinsey)

For marketers, that list should sound familiar.

The real blockers are usually not whether a team believes in responsible AI, but whether it has clear ownership, workable review processes, the right data practices, realistic budgets, legal coordination and enough operational discipline to turn principles into repeatable decisions. (Accenture)

What responsible AI means

Definitions vary but I’m sharing Accenture’s one: “the practice of designing, building and deploying AI in a manner that empowers employees and businesses and fairly impacts customers and society.” (Accenture)

If you want to create a marketing-friendly interpretation, that would be: "use AI to improve performance and creativity, but do it in ways that respect people, make interactions understandable, protect data, keep systems safe and dependable, and preserve accountability for outcomes."

Why marketing is especially vulnerable

Marketing is one of the earliest and fastest adopters of generative AI. The use cases are obvious: campaign ideation, copy generation, translation, image variation, audience insights, chatbot interactions, personalisation, and content operations.

The AI Index maps responsible AI across nine dimensions: fairness and bias, privacy, transparency and auditability, explainability, factuality and truthfulness, safety, robustness, accountability and human oversight. Those dimensions map directly onto marketing realities. A personalisation model can create unfair outcomes. A chatbot can mislead or hallucinate. An image generator can produce biased or brand-damaging outputs. A targeting workflow can cross privacy or consent boundaries. (Stanford HAI)

The cost of scaling without responsibility

When organisations scale AI without responsible foundations, the costs show up in four places.

• Brand risk. Harmful, inaccurate, discriminatory or manipulative outputs spread quickly when AI is embedded into public content and customer experiences.
• Trust risk. When AI-generated content or decisions reach customers without adequate review, brands may face inaccurate product claims or biased targeting: these damage the trust that marketing teams spend years building.
• Operating risk. Without clear governance, teams improvise around unclear ownership, conflicting priorities and weak escalation paths.
• Strategic risk. Responsible AI is increasingly shaped by formal governance expectations, eg: OECD AI Principles and ISO/IEC 42001 as growing influences on enterprise practice.

Responsible use creates measurable value

The World Economic Forum and Accenture frame responsible AI as a differentiator: it lets innovation scale safely, sustainably and inclusively. (World Economic Forum)

According to Accenture, organisations that put the right structures in place from the start (including responsible AI) achieve nearly three times the return on their AI investments compared with those that do not. (Accenture)

The AI Index adds a practical signal: more organisations report positive impacts from responsible AI policies on business outcomes, operations and customer trust, alongside reductions in incidents. (Stanford HAI)

Let me share four shifts happening for marketing leaders to be aware of:

• From tools to capabilities. The question is no longer which generative AI tools a marketing team should test. It is which capabilities the team needs to use those tools responsibly at scale: governance, data controls, review workflows, measurement, escalation.
• From speed to trusted speed. Delivering faster still matters. But speed without safeguards only makes failures travel faster. Responsible AI turns speed into something more durable: the ability to launch quickly without constantly trading away trust.
• From isolated experiments to governed systems. Many teams begin with scattered pilots and enthusiastic individuals. The next phase: classify use cases by risk, document ownership, define review points and make oversight part of the workflow.
• From compliance to brand promise. In customer-facing functions, responsible AI should not live only in policy documents. It should become part of how a brand communicates, personalises and runs automated workflows, without crossing lines that damage credibility or disrespect customers.

A four-step roadmap for marketing leaders

Accenture defines four pillars of Responsible AI: organisational, operational, technical and reputational. Below, I’ve laid out steps in each area to scale AI responsibly, framed around marketing leadership decisions.

1. Organisational. Name the owner and set intent

• Appoint a Responsible AI owner within marketing (often marketing ops).
• Clarify decision rights and align your Responsible AI stance with existing brand and company values.

2. Operational. Embed governance into workflows

• Add an AI review step into campaign, content and chatbot workflows whenever AI is used.
• Use a short checklist (fairness, privacy, transparency, factuality, brand safety) instead of long policies no one reads.

3. Technical. Choose tools and data with guardrails

• Prefer platforms with logging, access control and safety filters.
• Define rules for which customer data can be used with external models and how outputs are monitored over time.

4. Reputational. Measure and communicate trust impacts

• Decide what your brand will and will not do with AI (e.g., no undisclosed deepfakes or synthetic endorsements).
• Track complaints, escalations, opt-outs and corrections alongside campaign performance metrics.

Start with these before it is too late.

Next two years? The organisations that win over the next two years will be the ones that build the discipline to use the tools in ways customers can trust and teams can sustain.

Share with one marketer still scaling AI without a responsible-by-design approach.

← All insights