When AI Agents Become the Gatekeepers of Crowdfunding, Who Verifies the Evidence?

· By Michael Eis · Methodology

AI could fundamentally change how retail investors discover, analyse and invest in private companies. But as investment decisions increasingly move through intelligent agents, a more fundamental question emerges: who can investors trust to verify the information behind those decisions?

The next major transformation of equity crowdfunding may have little to do with crowdfunding platforms themselves.

For more than a decade, the industry has focused on making private investment opportunities accessible to ordinary investors. Platforms have simplified onboarding, digitised investment processes and created new channels through which startups and growing businesses can raise capital.

Yet the underlying investment journey has remained remarkably consistent. Investors discover an opportunity, visit a platform, read the pitch, review the available disclosures and decide whether to invest. Some conduct additional research. Many rely heavily on information prepared by the company seeking their money.

Artificial intelligence is beginning to challenge that model.

In the third and final episode of the Global Equity Crowdfunding Alliance's *Future of Crowdfunding* series, industry representatives explored what could fundamentally change crowdfunding over the next decade. Their discussion covered AI, tokenisation, institutional capital, investor behaviour and cross-border investment.

One of the most interesting possibilities was raised by Constantin from LenderKit: investors may eventually stop navigating crowdfunding platforms in the way they do today. Instead, they could ask an AI agent to identify suitable opportunities, compare investments and guide them through the investment process.

It is a plausible development. But it introduces a problem that deserves considerably more attention than the technology itself.

**When AI becomes the interface between investors and investment opportunities, who verifies the evidence on which those agents rely?**

From crowdfunding platforms to AI-mediated investing

Consider an investor interested in European renewable energy startups.

Today, that investor might visit several crowdfunding platforms, compare available campaigns, download Key Investment Information Sheets (KIIS), research the companies and assess the risks. The process is fragmented, time-consuming and frequently dependent on the investor's ability to interpret financial and commercial information.

Tomorrow, the investor might simply instruct an AI agent to find five European renewable energy investments, compare their valuations, examine their financial assumptions, identify material risks and present the most relevant opportunities.

The agent could search multiple platforms, retrieve disclosure documents, examine public company information and produce a comparative analysis in minutes.

For investors, this would represent a considerable improvement in access to information. For crowdfunding platforms, it could fundamentally change how investment opportunities are distributed and discovered.

As several participants in the GECA discussion observed, platforms and issuers may increasingly need to make their information accessible to AI agents rather than designing their entire information journey around human visitors.

That has important implications for the future of investment disclosures.

A PDF designed to satisfy regulatory requirements and inform a human reader is not necessarily an effective source for automated, comparative investment analysis. Machine-readable information, consistent data structures, document identification and reliable version histories could become increasingly important.

But making information accessible to AI does not make it accurate.

And making it accurate at the time of publication does not guarantee that it remains accurate throughout a fundraising campaign.

Faster analysis does not necessarily mean better due diligence

One of the most frequently discussed benefits of AI is its ability to reduce the time required to conduct due diligence.

That benefit is real. AI can help investors examine financial statements, identify inconsistencies between documents, compare assumptions, summarise commercial risks and locate information requiring further investigation.

It can also give retail investors access to analytical capabilities that previously required substantial time, specialist knowledge or professional assistance.

However, faster processing should not be confused with independent verification.

An AI system can analyse a company's revenue projections in considerable detail without establishing whether the underlying commercial assumptions are credible. It can summarise management's market assessment without discovering that important competitors were omitted. It can compare valuation multiples without questioning whether the selected peer companies are appropriate.

And it can produce a convincing investment analysis based almost entirely on information supplied by the company raising capital.

This distinction matters.

The regulatory disclosure framework for European crowdfunding, including the KIIS required under Regulation (EU) 2020/1503, provides a structured starting point for investors. But disclosure requirements do not automatically establish the independent accuracy of every commercial claim, projection or valuation presented to investors.

Under the European Crowdfunding Service Providers Regulation (ECSPR), crowdfunding service providers must undertake specified due diligence checks on project owners. Those minimum checks should not be confused with a comprehensive, independently verified investment assessment of every claim made in a fundraising campaign.

AI cannot resolve that distinction merely by reading the available documents more efficiently.

If multiple AI agents analyse the same incomplete or misleading information, they may simply reproduce the same weaknesses at greater speed and scale.

The risk is not merely that investors receive incorrect information. It is that AI-generated analysis can create the appearance of independent scrutiny where little independent verification has actually occurred.

The missing layer: evidence, provenance and independent scrutiny

The GECA discussion introduced another particularly important idea.

Neera Patel argued that AI-supported due diligence could become considerably more valuable when combined with verifiable, portable records of evidence. Rather than repeatedly performing the same checks, platforms and institutional investors could potentially exchange structured due diligence packages containing underlying documents, verification records, approvals and audit histories.

The argument is compelling, although blockchain is not necessarily required to achieve it.

The underlying requirement is a trustworthy record of what information was available, where it originated, when it was obtained, whether it changed and how particular analytical conclusions were reached.

That requires distinguishing between different types of information.

An issuer's statement about its commercial prospects is not equivalent to independently verified revenue. A market estimate prepared by management is not equivalent to external market evidence. An AI-generated finding is not equivalent to a conclusion reviewed and accepted by a qualified human analyst.

Nor does a document's presence in a secure or immutable repository establish that its original contents were truthful.

A cryptographic hash can help demonstrate that a particular document has not changed. It cannot demonstrate that the document was accurate when created.

This is an important limitation, particularly when discussing blockchain-based evidence systems. Technical integrity and evidential reliability are related, but they are not interchangeable.

A meaningful evidence framework should therefore distinguish between source integrity, source reliability, analytical interpretation and human verification.

It should also preserve conflicting evidence rather than forcing every available source into a single apparently consistent narrative.

That would create a more useful foundation for AI-assisted investment analysis than simply generating increasingly sophisticated summaries of issuer disclosures.

Who pays for independent verification?

There is, however, an uncomfortable question that rarely receives sufficient attention in discussions about improving investor protection.

Who is going to pay for independent due diligence?

Crowdfunding platforms generally operate commercial businesses built around attracting issuers, facilitating successful fundraising and generating revenue from their services. Their incentives are not necessarily identical to those of the investors providing the capital.

That does not make platforms inherently untrustworthy. But it creates an economic tension that should not be ignored.

The company raising capital wants its investment proposition presented as favourably as possible. The platform has an interest in attracting investment activity and facilitating successful campaigns. The investor needs an assessment that is prepared to challenge the proposition, identify weaknesses and potentially conclude that an investment should be avoided.

These interests can coexist, but they are not always aligned.

If an issuer pays directly for an independent assessment, questions arise about the analyst's commercial dependence on the company being assessed. If a crowdfunding platform commissions the analysis, similar questions may arise about the platform's influence over the assessment, particularly when negative conclusions could affect the success of a fundraising campaign.

And if the investor must pay for every individual due diligence report, the cost may become disproportionate to the relatively small investment amounts that characterise much of retail crowdfunding.

There is no perfect funding model.

An investor-funded service may offer a clearer alignment with investor interests, but it faces the practical challenge that many retail investors are unwilling to pay significant fees for research, particularly before deciding whether to invest at all.

A platform-funded model could make independent analysis available to a much wider audience. However, commercial arrangements would need to protect the analyst's independence, including the right to publish negative findings and maintain conclusions that the platform or issuer would prefer to challenge.

An issuer-funded model is not necessarily incompatible with independent verification either, provided the fee is not contingent on the result, the funding relationship is disclosed, and the methodology and editorial independence are protected.

A hybrid model may ultimately prove more sustainable, combining investor access, transparent research fees, platform integrations and clearly defined governance arrangements.

The central principle should be that whoever finances the assessment must not be able to purchase its conclusion.

This is particularly important as AI reduces the cost of producing investment analysis. Generating a detailed report may become inexpensive, but obtaining reliable evidence, resolving contradictory information, verifying important claims and accepting responsibility for analytical conclusions still require resources.

There is also a collective-benefit problem. Once independent research becomes publicly available, investors may benefit from it without contributing to its cost. That is good for information accessibility, but it complicates the economics of producing and maintaining the research in the first place.

The industry should therefore distinguish between the cost of generating an AI report and the cost of establishing credible independent evidence.

Those are not the same product.

And if crowdfunding is to become a substantially larger and more mature investment market, funding independent scrutiny cannot remain an afterthought.

Investor sovereignty: the principle we should not abandon

The most thought-provoking contribution to the GECA discussion came from Carsten Wenzlaff, who questioned whether the industry's enthusiasm for AI agents might undermine one of crowdfunding's original principles.

Equity crowdfunding emerged partly from the ambition to give individuals greater control over where they invest their money. Instead of relying exclusively on traditional financial intermediaries, investors could participate directly in financing companies and projects they believed in.

But what happens if those same investors increasingly delegate their investment decisions to AI systems?

An investor asking an agent to compare several investment opportunities is clearly using technology to support an individual decision.

An investor allowing an agent to select investments, determine allocations and execute transactions without meaningful oversight is doing something fundamentally different.

The distinction is not merely philosophical. Depending on the service and its implementation, automated recommendations, portfolio management and execution can raise additional regulatory questions under financial-services legislation.

There is also the question of independence.

An AI agent might prioritise investment opportunities according to risk, sector, expected returns or an investor's stated preferences. But it might also be influenced by the availability of data, commercial relationships, platform integrations or the information that issuers have optimised for AI discovery.

A system that appears to provide impartial investment selection may therefore introduce new forms of information asymmetry and conflicts of interest.

And the funding question becomes relevant here as well.

If an AI agent receives compensation for directing investors towards particular offerings, or if certain issuers receive preferential visibility through commercial arrangements, how can investors distinguish genuinely relevant opportunities from commercially promoted ones?

This is not entirely different from the conflicts already found in advertising, financial distribution and investment research. But AI agents could make those conflicts less visible, particularly when recommendations are presented through personalised conversations rather than conventional advertisements.

An agent that knows an investor's preferences, financial circumstances and previous investment behaviour may become extraordinarily effective at influencing investment decisions.

That capability creates opportunities for better investor support. It also makes transparency about commercial incentives and the selection of information essential.

This is where investor sovereignty becomes relevant again.

The objective should not be to prevent investors from using AI. It should be to ensure that investors understand the basis of the analysis, can question its assumptions, examine the supporting evidence and retain meaningful control over their decisions.

AI should expand investors' ability to exercise independent judgement, not quietly replace that judgement with an opaque automated process.

Regulation matters, but trust cannot be reduced to compliance

The GECA panel also discussed the EU AI Act and the increasing importance of governance, transparency and auditability.

These are legitimate concerns, but the regulatory distinction is important.

Not every AI system used by a crowdfunding platform or an investor automatically qualifies as a high-risk system under the AI Act. Classification depends on the system's intended purpose and the applicable legal criteria, not simply on whether it operates within financial services.

Under the amended EU AI Act implementation timetable, the principal high-risk requirements for systems covered by Annex III apply from 2 December 2027, while certain product-related high-risk requirements apply from 2 August 2028.

Where the AI Act's high-risk obligations apply, they include requirements concerning risk management, documentation, logging, human oversight and other safeguards. They should not, however, be interpreted as a blanket obligation for every crowdfunding platform to record every interaction with every AI tool.

Moreover, the AI Act is not the only relevant framework. Depending on the activity and operating model, GDPR, financial-services regulation, outsourcing arrangements, cybersecurity obligations and consumer or investor protection requirements may also be relevant.

But compliance alone will not solve the underlying problem.

An AI system may be legally compliant while producing an investment analysis based on weak or incomplete evidence. A crowdfunding platform may satisfy its disclosure obligations while investors remain unable to verify important claims independently.

The market therefore needs more than compliant AI.

It needs better standards for how evidence is collected, evaluated, challenged, documented and communicated.

And it needs commercial and governance structures that make independent scrutiny economically viable without compromising its integrity.

Where CrowdDiligence fits into this emerging market

At CrowdDiligence, our starting point is that investment disclosures should be the beginning of due diligence, not the end of it.

The KIIS and other issuer-provided materials are important sources. But meaningful investment analysis must also examine external evidence, question assumptions and distinguish verified information from statements that remain unsubstantiated.

CrowdDiligence is developing an independent due diligence environment built around structured findings, source evaluation, documented assessments and collaboration between human contributors and AI-supported analysis.

An important part of this approach is maintaining distinctions between AI-generated material, human-reviewed findings and the underlying evidence. Conflict-of-interest declarations, review responsibilities and auditability are also essential elements of a credible analytical process.

We do not consider AI-generated analysis to be independently verified simply because it has been produced by a sophisticated model.

Nor should the quality of an investment opportunity be determined by how convincingly an AI agent can explain it.

This approach becomes particularly relevant if the investment market moves towards AI-mediated discovery and comparison.

CrowdDiligence's longer-term opportunity is not simply to provide investors with another AI assistant. It is to help establish an independent analytical and evidential layer that investors, analysts and potentially their AI agents can use to examine private investment opportunities.

Such an infrastructure would need to support reliable source references, document versioning, provenance, clearly identified verification status and transparent analytical conclusions.

Several of these capabilities are already part of CrowdDiligence's developing architecture, while others remain important areas for further development.

The distinction matters. Building a credible evidence framework requires demonstrating what has been implemented, rather than presenting a future ambition as an established capability.

The economic model matters too.

If independent due diligence is to become an accessible part of the investment process, rather than a premium service reserved for larger investors, the industry will need funding arrangements that support reliable analysis while protecting its independence.

For CrowdDiligence, that is not simply a commercial question. It is part of the governance challenge that any credible independent due diligence infrastructure must address.

The next decade will not be decided by AI alone

The GECA discussion concluded by asking what would have to change for crowdfunding to become ten times larger over the next decade.

The participants offered several answers: greater investor participation, successful exits, portfolio diversification, international collaboration, institutional capital and more efficient technology.

All are relevant.

But there is a risk of overlooking a more fundamental constraint.

A market cannot build sustainable investor confidence simply by making it easier to discover and purchase investments. Investors must also be able to understand what they are buying, assess the associated risks and distinguish credible investment propositions from weak ones.

If AI agents make private investment opportunities dramatically easier to access without improving the quality and independence of the underlying evidence, the industry may succeed in scaling participation while simultaneously scaling its existing weaknesses.

That would be technological progress without a corresponding improvement in investor protection.

The more promising outcome is different.

AI could make independent due diligence more accessible, reduce unnecessary research costs, support portfolio comparisons and expose inconsistencies that individual investors might otherwise overlook.

It could also encourage platforms and issuers to publish more structured, verifiable and reusable investment information.

But this will require an infrastructure that treats evidence quality, provenance and independent judgement as essential components of digital capital markets, rather than optional additions to increasingly automated investment journeys.

It will also require answering the question of who pays for that infrastructure, who governs it and how its independence can be demonstrated.

Those questions may ultimately prove more important than which AI model performs the analysis or which blockchain stores the records.

Crowdfunding originally challenged the idea that access to private investments should be controlled by a limited group of financial intermediaries.

As AI agents become more influential, the industry should be careful not to replace those traditional intermediaries with a new generation of opaque digital gatekeepers.

The real opportunity is not to let AI decide what investors should believe.

It is to make the evidence behind investment decisions easier to find, easier to challenge and considerably harder to misrepresent.

**The future of crowdfunding should not be about delegating investor sovereignty to artificial intelligence. It should be about using artificial intelligence to make that sovereignty more meaningful.**

---

*This article was inspired by the Global Equity Crowdfunding Alliance's third Future of Crowdfunding roundtable, “Who Decides: AI Agents, Investor Sovereignty & the Next Decade”. The analysis and conclusions are CrowdDiligence's own.*

References

Tags: AI Agents, Equity Crowdfunding, Investor Sovereignty, Independent Due Diligence, Evidence Verification, AI Governance, Investor Protection, GECA