The impact finance sector is currently navigating a challenging period characterised by increased pressure to achieve greater results with fewer resources. In this environment, artificial intelligence (AI) has emerged as a critical productivity tool, though its actual efficacy and ethical implementation remain subjects of intense debate.

Smarter data, cheaper microfinance

Kicking off with a practical case, Yossef Zahar, CEO and fund manager at PandanGreen, highlighted AI’s transformative role in structuring securitisation vehicles and microfinance initiatives. His firm is launching a €150 million bond on the Luxembourg Green Exchange (LGX), with plans to reach €750 million within 30 months to finance five-year loans. It is targeting microfinance and non-banking financial institutions (NBFIs) with climate change solutions in the Asia-Pacific region.

Zahar emphasised that for these entities, AI significantly reduces operational costs by automating the creation of policies and the analysis of tens of thousands of data points related to ESG and gender impact. “We already budgeted €250,000 for a monitoring data tool,” he said.

With AI, Zahar noted that he can analyse ESG data patterns and errors more quickly and act faster to correct them. As operational costs decline, the fund can lower management fees. Consequently, these savings can be passed on to micro-entrepreneurs through improved interest rates.

AI keeping green claims in check

Similarly, EY has adopted an “EY Client Zero” strategy, testing AI solutions internally before deployment. Anna Illarionova, senior manager at EY Climate Change and Sustainability Services, explained that tools such as “AI Green Side” allow for the screening of green claims to ensure that fund managers do not overpromise on sustainable development goals or present inconsistent commitments.

AI hype meets harsh legal reality

Despite the enthusiasm for AI, legal expert Faustine Cachera, counsel at Arendt, warned of a “lack of adequate consideration for governance and data protection.” A significant concern involves the use of open tools such as ChatGPT, which may result in confidential client data being transferred outside the European Union, potentially breaching the General Data Protection Regulation (GDPR) or professional secrecy.

“If I put my client’s data on ChatGPT, I would be breaching attorney-client privilege, which is subject to criminal sanctions,” noted Cachera. “Sam Altman, before a US Senate committee hearing, openly said that ChatGPT is not able to abide by 50 different sets of state laws,” reported a participant in the audience.

Cachera stressed that even with paid licences, such as Microsoft Copilot, data remains subject to the US Clarifying Lawful Overseas Use of Data (CLOUD) Act, which may allow authorities to access servers located in Europe. “Microsoft says that it can indemnify you if it happens.”

“You cannot outsource your confidentiality obligations to your AI provider,” stated Cachera. To mitigate these risks, she advised organisations to integrate AI considerations into their existing governance structures, assign clear roles and responsibilities, and comply with applicable legal frameworks such as the Digital Operational Resilience Act (DORA).

Cachera went on to advise firms to establish robust AI acceptable-use policies; conduct comprehensive risk assessments of proposed AI use cases; carry out thorough vendor due diligence; put in place security measures on data; and invest in AI literacy for all employees.

AI’s next leap tests global rulebooks

There was also a debate over whether regulation can keep pace with technological advancement, particularly as experts predict the arrival of artificial general intelligence (AGI) within the next decade. “It will change everything in a much more fundamental way than the industrial revolution,” said one attendee.

Cachera argued that European regulations such as the AI Act protect fundamental rights and may eventually be adopted globally, like the GDPR. However, other audience members worried that strict frameworks might cause Europe to lag behind the US and China in the global AI race.

Alternative solutions may be more efficient

As organisations negotiate terms and conditions with large US tech firms, some are turning to “vertical AI”—specialised artificial intelligence systems or firms designed to solve specific tasks or address unique challenges within a particular industry.

Under the contractual agreements between an AI firm and its clients, Cachera explained that these tools can be trained solely on internal, secure databases to maintain data integrity and prevent proprietary information from training external models.

Profile of a future leader

The shift towards AI automation raised questions about the future of junior roles and leadership. While administrative and repetitive tasks are being automated, the consensus among panellists and audience members was that AI will not replace the need for human oversight, interpersonal connection, critical thinking, and empathy. A lawyer in the audience said, “When I speak to clients, I feel like I have to understand what they're trying to say and not what they're actually saying.”

Referring to a meeting with a senior manager at a software company, Zahar explained that the role of that senior expert has become akin to a “chess master” who directs multiple AI agents to perform coding or analysis. However, he stressed that he retains ultimate accountability for the output.

Panellists emphasised that future recruitment will likely focus on soft skills and philosophical thinking, as these are necessary to navigate the ethical complexities and “hallucinations” associated with AI models.

The hidden environmental cost of AI

Beyond legal risks, the environmental impact of AI is a growing concern, said Illarionova. The high energy and water consumption required to cool data centres poses a contradiction for impact investors focused on climate solutions. Furthermore, there is a risk of an “AI gap” between companies with the financial resources to develop the most sophisticated tools and those that cannot.

Despite these challenges, the discussion concluded on a positive note, as Zahar highlighted AI’s potential to democratise information and provide small businesses in emerging markets with better access to expert advice and affordable credit. Ultimately, the panellists agreed that while AI can accelerate impact, it must be governed by a logic that considers social, environmental, and ethical consequences to remain sustainable.