Coca-Cola's AI Literary Campaign Cited a Book That Does Not Exist

What happened

Coca-Cola's "Classic" campaign was built around a flattering premise: the brand has always been woven into the fabric of literature. To demonstrate it, VML New York and WPP Open X deployed an AI system to scan books for references to Coca-Cola, then surface the most notable ones for use in advertising. The approach was confident enough that the campaign ran publicly. The problem was that at least one of the books the AI cited does not exist.

The ad featured what it presented as a quote from a 1967 work titled "Extreme Metaphors by J.G. Ballard." No such book exists. The actual source was a 2012 posthumous collection of interviews called "Extreme Metaphors: Selected Interviews with J.G. Ballard 1967-2008." The distinction matters beyond the wrong date: the words used were not Ballard's prose but a translation of spoken interview answers, repurposed by the campaign as literary writing. The AI conflated title, year, format, and voice, then delivered the result as a confirmed literary reference.

Coca-Cola stated that a manual review had taken place before the campaign ran. If it did, that review failed to catch a fabricated book title, a wrong publication year, and a fundamental mischaracterization of what the source material actually was. The gap between "a human looked at this" and "a human verified this against the original source" is exactly where the error survived. Using AI to scan thousands of documents for brand mentions is not inherently reckless; treating its citations as verified without independently checking them against primary sources is.

The reputational stakes run in two directions. For Ballard's estate and readers, the campaign misattributes invented prose to a writer whose actual voice has been carefully preserved in posthumous editorial work. For Coca-Cola, the error made the campaign's central claim its own liability: that the brand's presence in literary history is worth celebrating, illustrated with a literary reference that never existed. The campaign intended to honor a legacy and instead manufactured one.

AI-assisted content production has no built-in mechanism for distinguishing a plausible citation from an accurate one. The only check available is human verification against primary sources, conducted before publication, not after. Without a provable record of what the system generated, what a reviewer actually examined, and which source confirmed the output, there is no way to locate where the fabrication entered the pipeline and no basis for preventing the same pattern from recurring in the next campaign.

Reported impact

Affected parties

Not publicly disclosed

Harm type

Not publicly disclosed

Scale

Not publicly disclosed

Financial impact

Not publicly disclosed

Regulatory action

Not publicly disclosed

Classification

Organization

Not publicly disclosed

AI system

Not publicly disclosed

Industry

Not publicly disclosed

Country

Not publicly disclosed

Provider

Not publicly disclosed

Incident type

Not publicly disclosed

Relevant governance controls

Governance control mapping is not available for this record.

• No controls mappedNot publicly disclosed

Control mapping is analytical. It does not state that any control would have prevented the incident.

Sources and evidence

This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.