The threat posed by artificial intelligence is moving beyond traditional cyberattacks and misinformation. Businesses are now facing a growing problem: how to prove that a corporate message, executive statement, video or financial announcement is actually genuine.

Imagine watching a video in which a company's chief executive recommends an investment opportunity. The person's face looks real, their voice sounds familiar and the company's branding appears authentic. The message could easily convince customers or investors to act. The problem is that the executive may never have made the statement at all.

This type of deception is becoming increasingly difficult to identify as generative AI tools become more accessible and sophisticated. Fraudsters can now create realistic videos, replicate voices and produce convincing documents without requiring the specialist skills or resources that were previously necessary.

South African institutions have already encountered the problem. The South African Reserve Bank has warned about fabricated AI-generated videos featuring Governor Lesetja Kganyago, including fake material designed to make it appear as though he was involved in televised confrontations or was promoting investment opportunities.

The country's Financial Sector Conduct Authority has also issued warnings about scams that use deepfake videos and the identities of prominent South Africans to make fraudulent investment schemes appear legitimate. Public figures whose identities have been exploited include Springbok captain Siya Kolisi, President Cyril Ramaphosa and businessman Patrice Motsepe.

The problem is not limited to South Africa. Earlier this year, police in Hong Kong revealed that employees at a multinational company had been tricked into transferring more than US$25 million after criminals used AI-generated video and audio to impersonate senior executives during a video conference.

Such incidents demonstrate why deepfakes are becoming a serious business risk rather than simply an internet nuisance. A fake statement from a senior executive could potentially affect a company's reputation, influence investors or trigger financial losses before the organisation has enough time to issue a correction.

The World Economic Forum has also highlighted AI-enabled misinformation and disinformation as a major short-term global risk. The concern is that increasingly convincing synthetic content could weaken public confidence in businesses, financial institutions, governments and other organisations.

The problem with simply detecting deepfakes

For years, companies have invested in cybersecurity systems and crisis-communication strategies designed to respond to false information. However, AI-generated content creates a different challenge because the fabrication itself can look remarkably authentic.

A fake photograph, voice recording or corporate announcement can be produced and distributed within minutes. By the time an organisation discovers the content and publicly denies it, the original material may already have spread across social media, messaging applications and other online platforms.

This is why the conversation is increasingly shifting from simply detecting fake content to proving the origin of legitimate content.

One approach gaining attention is digital provenance. The concept is similar to provenance in the art world, where the history of an artwork—including its creator, ownership and restoration—is used to establish whether it is authentic.

For digital content, provenance systems can record information about where a photograph, video or document came from, how it was produced and whether it was subsequently modified.

Content Credentials could help establish authenticity

One of the major initiatives in this area is the Coalition for Content Provenance and Authenticity (C2PA). The open standard allows creators and organisations to attach cryptographically signed information to digital files.

This information can include details about when content was created, the device or software used to produce it and whether artificial intelligence played a role in its creation.

The technology is gaining support across the technology and media industries. Adobe has incorporated Content Credentials into its creative tools, while companies including Microsoft and Google are participating in efforts around digital provenance. OpenAI has also introduced Content Credentials for images generated through ChatGPT.

Major camera manufacturers, including Canon, Nikon, Leica and Sony, are developing technologies that can capture authenticity information when photographs are taken.

However, provenance is not a perfect defence. Metadata can sometimes be removed, and its effectiveness depends partly on digital platforms preserving and displaying the information attached to content.

That means businesses are unlikely to rely on one technology alone. Provenance may need to work alongside watermarking, deepfake detection systems, platform policies and greater public awareness of AI-generated content.

Authenticity becomes a business responsibility

For organisations, the challenge is becoming broader than simply responding to false information after it appears online. Companies may increasingly need systems that allow customers, journalists, investors and other stakeholders to verify that official communications genuinely came from the organisation.

Richard Frank, chief technology officer at Flow Communications, describes this as an emerging operational challenge for businesses. As synthetic content becomes harder to distinguish from genuine material, establishing the origin of corporate communications could become an important part of maintaining public trust.

The shift could fundamentally change how companies approach digital communication. In the past, organisations focused primarily on ensuring that their published information was accurate. In an environment flooded with convincing AI-generated fabrications, they may also need to demonstrate that their communications are authentic.

For businesses, that means the future of digital trust may not depend solely on what a message says or how convincing it looks. It may increasingly depend on whether the organisation can prove where the message came from and whether it has been altered.

As AI-generated content becomes more common, authenticity could become a competitive advantage—and organisations that can make their communications easier to verify may be better positioned to maintain trust with customers and investors.