AI & Ethics: Deploying a Bad AI Agent in a Sustainable Sector is a Reputational Hazard

Introduction

In a sector where credibility is as valuable as currency, a poorly built AI agent doesn’t just create operational problems, it exposes an organization to a potential loss of public trust and even legal liability. While agentic AI is still navigating its own reputational obstacles, it’s incumbent upon organizations in Renewable Energy and Sustainable Finance that are exploring AI to do so with an eye on transparency and integrity. One of the simplest ways to approach this mandate is to establish a Mission-Aligned AI Standard that guides every decision made during agentic AI adoption, development, and deployment.

---

Why Sustainable Organizations Must Take Care When Adopting AI

AI technology offers a lot of promise around efficiency and cost reduction. However, any sustainable company exploring the use of AI agents needs to do so with heightened awareness of potential negatives that could result from that choice, and also have a strategy in place to avoid them.

—

In June of 2023, several months after AI technology saw its first massive wave of public adoption, two attorneys working for a New York firm submitted a problematic legal brief to a federal court prepared with the use of ChatGPT. The brief cited more than a half dozen case law examples, none of which actually existed, because ChatGPT had fabricated every single one of them.

The two lawyers were sanctioned by U.S. District Judge Kevin Castel, fined, and publicly rebuked in a ruling that generated headlines across the world and placed a blemish on their professional careers. This was an embarrassing and highly visible faux pas, but because mass awareness of AI was still developing at the time, the attorney’s poor judgment could easily be marked up as an early adopter’s mistake.

Years later, awareness of AI, what it is and how it works, has grown considerably and public tolerance for such failures has arguably diminished. According to the 2025 Ipsos AI Monitor, a survey of more than 23,000 adults across 30 countries, 53% of respondents expressed feeling nervous about AI. In the United States, the Edelman Trust Barometer found that only 32% of Americans trust AI, down from 50% just five years ago.

Any sustainable organization considering using AI tech should be ready to do so in a transparent and demonstrative manner.

There is also credible concern about AI technology’s environmental impact. An expanding body of research is being produced by non-profits like the National Bureau of Economic Research (NBER) which brings awareness to the environmental costs of unchecked AI-driven data center sprawl.

While back in 2023 the attorneys’ poor judgment caused them embarrassment, a relatively small fine ($5,000 each) and a dent in their professional reputations, in today’s atmosphere there are higher stakes for any business or organization (especially sustainable ones) that wade into AI without due diligence. Specifically with Renewable Energy and Sustainable Finance, the environmental dimension adds an additional layer of accountability. Does this mean a sustainable organization should steer clear of AI? Not necessarily. Instead, these types of entities must earn the right to deploy it by clearly demonstrating that they are implementing AI responsibly.

It’s Not Just About Reputation – Legal and Financial Ramifications Are a Real Concern

Public embarrassment is certainly painful terrain for any brand to navigate, however, the impact from regulatory actions presents injury of a different sort. Such events can result in an organization losing access to government-backed funding, being buried under extraordinary fines, and tied up in expensive state and federal cases for years.

—

In an effort to make their shares more attractive to ESG (environmental, social, governance) investors, some entities may engage in “greenwashing,” a practice explicitly prohibited by the SEC. In 2023, DWS Investment Management Americas, the asset management arm of Deutsche Bank, agreed to pay $19 million to settle SEC charges for this exact violation.

In the case, the SEC determined that DWS failed to implement ESG integration policies it had publicly advertised. Additionally, its marketing claims about being an “ESG leader” and having sustainability in its “DNA” were found to not reflect reality. The organization’s troubles didn’t end there. In April 2025, German prosecutors concluded a three-year investigation into the firm by adding a €25 million ($27 million) fine on top of what the SEC had issued.

The final cost to DWS? $46 million in fines, a forced CEO resignation, police raids on company offices, and a reputational wound that made international headlines for four years.

While DWS didn’t deploy a faulty AI agent, the situation demonstrates a regulatory pattern worth understanding. An organization that makes unsubstantiated claims about an AI tool’s capabilities faces deliberate misrepresentation charges. However, a well-meaning organization that deploys a poorly designed agent that hallucinates unsubstantiated claims faces the same scrutiny, and as established in Moffatt v. Air Canada, the law does not accept rogue AI output as an excuse.

This is an emerging form of deception that regulators are now calling AI-washing. In March 2024, the SEC formally defined the term and acted on it simultaneously, filing charges against two investment advisory firms, Delphia and Global Predictions, for exaggerating the role of AI in their products. The firms collectively paid $400,000 in civil penalties. By January 2025, the SEC had extended similar charges to several publicly traded companies, including one listed on the Nasdaq.

In an evolving landscape where public and regulatory scrutiny is increasing on a global scale, greenwashing and AI-washing pose a convergent risk for sustainable organizations. An AI agent can either help a company move its mission forward, or expose an organization to public and regulatory harm. Part of ensuring that risk is properly managed involves committing to a Mission-Aligned AI Standard that guides how your business adopts, builds, and deploys AI systems.

What is a Mission-Aligned AI Standard?

Globally, AI regulation is moving from theory to enforcement; but, compared to firmly established industries like healthcare, aerospace, or public utilities, AI technology is still largely a self-governed space. In this atmosphere, navigating the risks outlined above requires more than good intentions. It’s far more proactive to have an organization-wide, mission-aligned framework that governs every decision made during AI development and deployment.

At Funnel Amp this collection of best practices is referred to as a Mission-Aligned AI Standard or MAAS. A MAAS is not a rigid compliance checklist. Instead, it’s a living framework, shaped by each organization’s mission statement, value proposition, sector, and operational context. A well prepared MAAS should be able to answer key questions such as:

• What specific operational problem will an AI agent solve?
• What evidence supports the decision to use AI rather than a human workflow?
• How will human accountability be incorporated into an agent’s operations?
• What steps have been or will be taken to minimize the AI agent’s environmental footprint?

While every organization’s MAAS will reflect its own principles, priorities, and existing regulatory obligations, the following elements form a practical foundation that any Renewable Energy or Sustainable Finance organization can adapt and build from.

Recommended elements for a Mission-Aligned AI Standard

1. Conduct a Reasonable Justification Analysis
   Before launching any AI project, document a clear business case explaining why AI is the right tool for the problem. A net-positive justification asks whether the operational benefit of deploying AI, measured in efficiency gains, emissions reductions, or mission outcomes, demonstrably outweighs any environmental or human costs.
   
   
2. Incorporate Human-in-the-Loop (HITL) Design
   Where operationally feasible, build human review into AI workflows at key decision points. HITL design is an architectural choice made during development; it ensures that consequential outputs are reviewed by a qualified person before triggering action.
   
   
3. Define Clear Human Accountability at Every Decision Point
   Separate from system design, every AI deployment needs a named human owner responsible for its outcomes. When an AI output affects a person, a community, or a regulatory filing, someone in the organization must be accountable for that result.
   
   
4. Establish a Knowledge Base Quality and Governance Policy
   A knowledge base is the critical layer in agentic AI that determines its data retrieval accuracy. The higher the quality, the more efficient. Define who owns the knowledge base, the standard by which it’s built and deployed, how frequently it is reviewed, and what operational or regulatory changes trigger an update. A knowledge base without governance becomes stale, and a stale knowledge base is a liability.
   
   
5. Prioritize Renewable Energy and SLM to Power AI Infrastructure
   Data centers are energy-intensive, and AI workloads amplify that demand significantly. Where possible, organizations should pursue local hosting powered by behind-the-meter renewable energy, or select cloud regions verified to use real-time renewable energy matching rather than carbon offsets. For routine and targeted tasks, Small Language Models (SLMs) consume significantly less energy than large generalist models (LLMs) and are often entirely sufficient for mission-specific applications.
   
   
6. Select LLM Vendors With a Clear Environmental Mandate
   If dependence on a cloud-based LLM provider is necessary, look beyond general sustainability pledges. Prioritize vendors who can demonstrate real-time renewable energy matching in the specific regions where your data is processed, not broad carbon offset purchases. Providers vary significantly in per-query emissions transparency, and that variance has real environmental consequences at scale.
   
   
7. Require Documented Carbon Reduction Practices From Outside Developers
   When engaging external contractors or consultants, treat environmental responsibility as a selection qualifier. Ask each vendor how they minimize the computational footprint of the systems they build, and whether their own operations reflect documented carbon reduction commitments. Contractors who cannot speak to either introduce reputational inconsistency into your supply chain.
   
   
8. Align AI Deployments With Applicable Regulations and Reporting Frameworks
   Each sector carries its own regulatory stack. Green Energy organizations may need to account for IRENA guidelines and grid safety standards; Sustainable Finance organizations operate under CSRD, ISSB, and SEC disclosure rules. A MAAS should explicitly map each AI deployment to the frameworks governing it, and flag when regulatory updates require a knowledge base review.

Ease Your Organization’s Adoption of AI with MAAS

AI presents many opportunities for sustainable organizations, but any company considering its adoption should be ready to do so in a transparent and demonstrative manner. Building a Mission-Aligned AI Standard that genuinely reflects your organization’s values, injects a degree of protection into your AI operations. It establishes a robust framework for present and future AI deployment, better positions you to accomplish mission goals, and publicly signals a commitment to responsible AI stewardship. As AI regulation evolves, these are all positives that will solidify a company’s status as an early supporter of responsible AI deployment.

---

References

1. Mata v. Avianca — Attorneys sanctioned for AI-generated false citations in federal court filing (CNBC, June 2023)
   https://www.cnbc.com/2023/06/22/judge-sanctions-lawyers-whose-ai-written-filing-contained-fake-citations.html
2. 2025 Ipsos AI Monitor — Global survey of public sentiment toward AI across 30 countries https://www.ipsos.com/en/conflicting-global-perceptions-around-ai-present-mixed-signals-brands
3. Edelman Trust Barometer Flash Poll — AI trust levels in the United States (November 2025) https://www.edelman.com/trust/2025/trust-barometer/flash-poll-trust-artifical-intelligence
4. National Bureau of Economic Research (NBER) — Environmental costs of AI-driven data center expansion https://www.nber.org/papers/w35100
5. DWS Investment Management Americas — SEC greenwashing settlement, $19 million (ESG Today, 2023) https://www.esgtoday.com/deutsche-banks-dws-fined-27-million-for-greenwashing/
6. DWS/Deutsche Bank — German prosecutors’ €25 million greenwashing fine (ESG Dive, April 2025) https://www.esgdive.com/news/german-prosecutors-slap-27m-greenwashing-fine-on-deutsche-bank-dws/744507/
7. Moffatt v. Air Canada — Civil tribunal ruling establishing organizational liability for AI chatbot errors (UBC Allard School of Law Review)
   https://commons.allard.ubc.ca/cgi/viewcontent.cgi?article=1376&context=ubclawreview
8. SEC AI-washing enforcement actions — Delphia and Global Predictions civil penalties (Latham & Watkins, March 2024)
   https://www.lw.com/admin/upload/SiteAttachments/SEC-Announces-First-Ever-Enforcement-Actions-for-AI-Washing.pdf