Lloyds' Marketing Chief: Our AI Videos Were Degrading the Brand

In June 2026, Lloyds Bank posted a series of AI-generated videos across its social channels starring a miniature version of its signature black horse mascot. In the clips, the tiny horse gets carried into an office in an employee's tote bag, drinks water from a glass, and knocks pens out of a pot, all built to promote the bank's pet insurance product. Four months later, at Marketing Week's Festival of Marketing on October 14, Lloyds' top marketer Suresh Balaji used the campaign as a cautionary case study in front of a room full of his peers, as reported by Marketing Week and corroborated by Campaign.

Balaji asked the audience directly whether they thought the videos qualified as "AI slop." As more hands went up, he gave his own verdict: "We very quickly realised that we were degrading our brand by creating this." It is a rare moment of a major financial brand naming its own AI misstep on stage, and it is worth more attention from marketing and branding teams than the usual AI trend roundup.

The Videos Tested Well, But Something Was Still Off

What makes the Lloyds example useful rather than just embarrassing is that it was not a flop by conventional metrics. Reporting indicates plenty of customers genuinely liked the videos; the clips were playful, on-brand in subject matter, and used a mascot people already recognized. The problem Balaji identified was not audience reaction in isolation. It was that the team's own instincts told them something did not line up, even while engagement looked fine.

The mismatch, as Balaji described it, was between the tone of AI-generated novelty content and what a 260-year-old bank brand is supposed to represent: stability, trust, and dependability. A tiny animated horse spilling a pen pot is charming in isolation, but it does not reinforce the qualities a bank needs a customer to associate with it when they are choosing who holds their money. Likes and shares measured attention. They did not measure whether the content was doing the brand's actual job.

Why This Matters Beyond One Bank

Lloyds has the budget, legal review, and brand governance of a major financial institution, and the videos still got made and published before anyone caught the disconnect. That is the real signal for smaller agencies, firms, and nonprofits experimenting with generative AI for social content: if a heavily resourced brand can greenlight something that quietly undercuts its own positioning, a leaner marketing team without a dedicated brand strategist in the room is at even higher risk of the same mistake.

Balaji's advice to the room was not to avoid AI. It was to use it "where it would genuinely make a difference," rather than reaching for it because a format is novel, cheap to produce, or likely to perform well in isolation on a feed. That is a useful filter for any team evaluating whether a piece of AI-generated content belongs in front of customers: does this reinforce what the brand stands for, or does it just reinforce that the brand has access to AI tools?

Engagement tells you people noticed. It does not tell you what they now believe about your brand.

A Quick Gut-Check Before AI Content Goes Live

Teams without a formal brand council can still borrow the discipline Lloyds applied after the fact, just earlier in the process. Before publishing AI-generated or AI-assisted creative, it is worth asking a short set of questions as a group, not just one person's gut call:

• Would this piece of content make sense coming from a human on our team, in our actual voice, or does it only work because it is AI-generated novelty?
• Does the tone match what we want a prospect to believe about us, or does it just match what performs well generally?
• If a competitor posted this exact piece, would it still feel distinctly like our brand, or is the mascot or logo doing all the identity work?
• Are we measuring success by engagement alone, or are we also checking it against our actual brand style guide and positioning?

None of these questions require banning AI tools. They require treating AI output the way you would treat a first draft from a junior creative: useful, fast, and still subject to the same brand judgment everything else goes through before it reaches a customer.

The Takeaway for Agencies and In-House Teams

The most striking part of Balaji's talk was not that Lloyds tried something that did not land. Brands test creative ideas constantly, and most tests are quietly retired without a postmortem on a conference stage. What stands out is that a major brand's own marketing leadership was willing to say, publicly, that a well-performing campaign still missed the mark on brand fit. That kind of self-correction is harder to pull off without someone in the room whose job is specifically to protect the brand's positioning, not just ship content.

If your team is experimenting with AI-generated video, imagery, or copy and nobody is explicitly responsible for asking whether it still sounds like you, that is a gap worth closing before the next campaign goes out, not after. Book a free consultation and we will help you build the brand guardrails that keep fast, AI-assisted production from quietly drifting away from who you actually are.