Artificial intelligence (AI) has been touted as playing various roles, from revolutionizing productivity to acting as almost autonomous decision-makers. However, as AI tools gradually integrate into daily operations, concerns are arising. People are starting to wonder: What will happen if the promises made by AI products fail to materialize? From exaggerated accuracy claims to opaque performance metrics, questions are being raised: How should AI systems be remediated when they underperform?
This disconnect has drawn the attention of consumer protection agencies, lawyers, and AI experts. They assert that marketing claims need to be substantiated with tangible enforcement. Last year, speculation surrounding breakthroughs in generative AI—an common form of the technology—led to what IBM’s Chief Inventor and UN AI Advisor Neil Sahota termed as “false marketing tactics.”
This phenomenon is known as “AI washing.” Similar to the concept of “greenwashing” where companies falsely label their products as environmentally friendly or sustainable, “AI washing” refers to companies making false or exaggerated claims about the capabilities of their AI models to appear more advanced, often to attract investment or gain a competitive edge in the market.
Subsequently, the Federal Trade Commission (FTC) in the United States has taken the lead in various initiatives to hold AI companies accountable for their products. In 2024, the agency launched “Operation AI Comply,” aimed at taking enforcement actions against companies leveraging “AI hype.”
The initiative seeks to combat behaviors of using AI tools to “deceive, mislead, or defraud people.” “The Federal Trade Commission is committed to ensuring that AI companies can innovate rapidly while maintaining the safety, reliability, and objectivity of AI platforms in line with the President’s AI Action Plan,” said Christopher Bissex, Deputy Director of Public Affairs at the FTC.
Bissex mentioned that some recent actions by the agency included issuing orders against companies making false claims about the accuracy or effectiveness of their AI products, suing companies for false marketing claims, and conducting research to gather information on how companies address potential risks posed by AI chatbots, especially in cases involving children.
Christopher Trocola, Founder of ARC Defense Systems, emphasized that a lack of accountability mechanisms is not due to inadequate regulation. He stated that the problem lies in the misunderstanding of how existing laws apply to AI, with companies believing they are not liable as there are no specific laws targeting AI. Drawing on his experience in the emerging industry, Trocola highlighted the need for an AI compliance framework to prevent potential pitfalls and protect consumer rights.
To ensure accountability and transparency in the realm of AI investments, experts suggest increasing auditing mechanisms and enhancing audit procedures as key elements for companies to verify whether they are being misled in their AI endeavors.
According to Chad Silver, Founder of Silver Tax Group, regulatory oversight surrounding the promises made by AI companies is proving to be deficient. Companies often make commitments but fail to provide audit records to prove their compliance when challenged.
MIT’s research project on “Networked AI Agents in Decentralized Architecture” disclosed that despite investments exceeding $40 billion in generative AI, 95% of companies did not realize any returns. This analysis, based on a systematic review of over 300 AI projects, interviews with 52 organizations, and feedback from 153 industry leaders, offers a sobering depiction of the AI landscape away from marketing hype.
Likewise, a RAND Corporation analysis revealed that over 80% of AI projects end in failure, twice the failure rate of non-AI IT projects. While AI supporters attribute failures to management’s lack of understanding of AI and skewed metrics for measuring post-deployment success, G2’s study showed that 57% of companies quickly transitioned AI agents from testing to formal production environments, indicating a short testing-to-scaling cycle.
Furthermore, Boston Consulting Group’s study found that only 5% of companies globally managed to achieve significant value from AI at scale. Notably, 60% of enterprises, despite substantial investments, either failed to create any tangible value or generated minimal returns.
Experts stress the importance of enhancing transparency and strengthening audit mechanisms as crucial ways for companies to ascertain whether they are being misled in their AI ventures.
Trusted audits should include checks at the model level, security testing, source traceability reviews, and assessments of organizational governance, according to Gourley. He added that audits should be conducted regularly, with some details publicly available while specific internal issues remain confidential as business secrets.
Trocola emphasized the need for a shift in AI audit methodologies, focusing on evaluating AI security rather than monitoring the employees using the technology. He pointed out that real scrutiny should target AI security issues such as AI drift, where AI model performance and accuracy degrade over time, potentially leading to hallucinations, data leaks, biases, and other problems.
Third-party performance audits are seen as essential by Silver, who underscored the need for independent organizations to conduct audits every six months to review model weights and biases in a credible manner.
As calls for greater transparency in AI investments grow louder, the push for standardized documentation requirements, data sources, limitations, risk analyses, and mitigation plans is seen as critical to each high-impact area, according to Gourley. He advocates for certification efforts directed by hybrid organizations or US public entities like the National Institute of Standards and Technology, or in collaboration with AI regulatory bodies and independent experts.
In December 2025, the White House issued an executive order on the “Ensuring AI National Policy Framework,” pledging to establish a minimally burdensome regulatory structure, including an AI litigation task force, evaluations of state AI laws, and potential federal reporting and disclosure standards.
Stanford University’s Human-Centered Artificial Intelligence Institute researchers predicted that 2026, after years of investments totaling billions of dollars, could mark a watershed moment for AI, where the practical utility of AI will face strict scrutiny. They emphasized the need for rigor over hype in the coming year, shifting the focus from “Can AI do it?” to “How is it done? What are the costs? Who does it serve?”
The World Economic Forum stated in December 2025 that if 2025 was the year of “AI hype,” then 2026 may be the year of “AI reckoning.” This signifies a critical period where the actual outcomes of AI technologies are expected to be reckoned with, shedding light on the realistic landscape of AI applications and investments to drive responsible innovation in the industry.