Atlas warns of AI reputation breach for businesses

Atlas Digital has reported widespread inaccuracies in how businesses are represented on major artificial intelligence platforms, with audits finding the problem across SaaS, technology and financial services brands.

The findings point to what Atlas calls an “AI reputation breach”, where services such as ChatGPT, Google Gemini, Perplexity AI and Claude generate information about companies that is incorrect, outdated or misleading. In Atlas's review, 72 per cent of brands had at least one factual error in AI-generated responses, while 70 per cent did not appear in AI recommendations for their own category.

The research adds to concerns that AI tools are becoming a new layer of public-facing information without the checks businesses would normally expect from search listings, company profiles or media coverage. According to Atlas, this can affect reputation and, in some cases, compliance, without a business being aware of the issue.

One factor heightening the risk is user behaviour. More than 80 per cent of people do not check the accuracy of AI-generated information, according to data cited by Atlas, meaning incorrect answers can influence decisions without further verification.

Audit findings

The issue was especially pronounced among businesses with more complex products or services, and among companies that had recently changed their branding or market positioning. In one example from the analysis, a technology company had just 44 per cent accuracy in AI-generated descriptions of its business.

That company was represented across 31 third-party platforms, and Atlas found outdated or incorrect information on every one of them. In another case involving a financial services firm, AI-generated descriptions included fabricated policies, incorrect products and fees that did not exist.

Those examples reflect a broader concern for sectors where precision matters. Financial services groups, software providers and other companies selling specialist products often rely on clear, current descriptions of what they do, what they charge and how they differ from rivals.

Ryan McMillan, Founder and Chief Executive Officer of Atlas Digital, said the shift in consumer behaviour has raised the commercial stakes for businesses that are not monitoring AI-generated results.

“AI is now a front door to decision-making, but it's not always getting the facts right,” McMillan said.

“There's no alert, no complaint. The customer simply chooses a competitor, and the impact goes unseen.”

Growing influence

Atlas argues the commercial effect of these errors is increasing as AI systems become more widely used in search and product discovery. ChatGPT reaches more than 900 million weekly users, while Google's AI Overviews are seen by more than two billion people globally, according to figures cited by the company.

Atlas also pointed to rising traffic from AI services to company websites in Australia, saying referral traffic has climbed 1,200 per cent year on year. Nearly half of consumers now use AI to inform purchasing decisions, based on the figures it cited.

Alla Lvovich, Organic Product Lead at Atlas Digital, said the link between AI answers and revenue is becoming clearer for businesses.

“ChatGPT alone now reaches more than 900 million weekly users, while Google's AI Overviews are seen by more than two billion people globally. In Australia, AI referral traffic to websites has increased by 1,200 per cent year on year.

“Nearly half of consumers now use AI to inform purchasing decisions, and that behaviour is translating into revenue. AI-driven traffic converts at between 12 and 17 per cent, compared with around 3 per cent for traditional search, meaning inaccuracies are more likely to directly impact sales,” she said.

Atlas also cited hallucination rates across major AI models ranging from 15 to 52 per cent, with false claims sometimes repeated and reinforced over time. This can create persistent errors in how a business is described, particularly when incorrect information is picked up across multiple sources.

Corporate response

Businesses are beginning to treat the issue as more than a marketing concern. Atlas said 72 per cent of S&P 500 companies now disclose AI-related risks in annual reports, up from 12 per cent in 2023, suggesting boards and investors are paying closer attention to misinformation generated by AI systems.

McMillan said companies should treat AI visibility and accuracy as a central commercial issue rather than a secondary communications task.

“This is where cybersecurity was a decade ago. It's not just a marketing issue, it's a genuine business risk,” McMillan said.

“But for companies that get it right, it's also a competitive advantage. If AI is shaping customer decisions, then actively managing how your business is represented becomes critical.”