Why AI Transparency Can Complicate Crowdfunding

Generative artificial intelligence can help a small team produce campaign images, draft copy, and develop prototypes with fewer resources. However, the commercial value of those efficiencies depends on whether customers still believe in the people and products behind them. A cheaper production process offers limited advantage if it makes potential supporters hesitate.

A study published in Decision Support Systems, “Will your AI turn backers away? Evidence from Kickstarter’s AI-use disclosure policy,” examines that tension through actual crowdfunding outcomes.¹ Its findings suggest that making AI involvement visible can coincide with a substantial funding disadvantage, even after researchers account for several differences between campaigns.

For creators, platforms, and investors, the opportunity is to understand what disclosure leaves unanswered. Identifying AI use tells an audience something about production. It does not necessarily explain who made the important decisions, whether a prototype works, or who will deliver the promised product.

What the Kickstarter AI Disclosure Study Found

Researchers Sonika Singhal, Yeongin Kim, Seonjun Kang, and Victoria Yoon analyzed 63,512 Kickstarter campaigns launched between Januar 2022 and Dezember 2024. The period spans the introduction of Kickstarter’s AI disclosure requirement in August 2023.

The researchers identified disclosed AI use through the presence of an explicit “Use of AI” section. They measured campaign success by whether pledged funding reached the goal and examined backer participation, funding targets, language characteristics, and creator experience.

They also used propensity-score matching to compare disclosing campaigns with campaigns that had similar observed characteristics. The matched sample contained 8,959 campaigns. Across their main specifications, disclosed AI use was associated with lower success.

In the fully adjusted matched model, the estimate corresponds to approximately 50% lower odds of reaching the funding goal. That is not a 50% reduction in success probability. For illustration, halving the odds associated with a 60% probability produces a probability of about 43%, not 30%. This example explains the statistical distinction; it is not a predicted outcome for a particular campaign.

The participation finding matters. The funding gap was associated with fewer people backing projects, rather than creators consistently asking for more money. That directs attention toward audience evaluation and willingness to participate.

Why AI Labels Can Create a Credibility Gap

Crowdfunding asks people to support something that often does not yet exist in finished form. Backers must assess both the proposed product and the creator’s ability to deliver it. Campaign materials therefore serve as evidence of preparation, competence, and commitment.

AI complicates that assessment because polished presentation may require less effort than it once did. A convincing image could depict a working prototype, a design concept, or an entirely generated scene. The same visual quality can represent very different levels of progress.

This suggests an economic consequence beyond the paper’s measured results: as attractive presentation becomes easier to produce, verifiable execution may become more valuable. Demonstrations, development records, and previous deliveries can help audiences distinguish projects that otherwise look equally impressive.

The study’s stronger negative association in technology campaigns is especially interesting. An audience interested in innovation may still require evidence of engineering capability. Enthusiasm for technology does not remove concerns about whether an unfamiliar team can build and support a product.

This reasoning also connects with Securities.io’s recent coverage of employee trust and automation ROI. Different stakeholders make different decisions, but technical capability must still translate into human acceptance before its full economic value can emerge.

Better AI Transparency Requires More Context

The findings should not encourage creators to hide AI involvement. Kickstarter’s AI disclosure policy established transparency requirements without banning AI projects. Concealment would also leave the underlying questions about authorship and delivery unresolved.

A more useful approach is to explain the scope of AI involvement alongside evidence of human responsibility. Consider two hypothetical campaigns that both disclose AI-generated images. One uses them to illustrate a product already demonstrated in a working video. The other offers generated renderings as its main evidence of feasibility. An identical label cannot communicate that difference.

Creators can make disclosure more informative by answering three questions:

• Which parts of the project used AI, and for what purpose?
• Who reviewed the outputs and remains responsible for delivery?
• What evidence demonstrates progress beyond generated materials?

These are proposed communication practices, not interventions proven effective by this study. The observed relationship with readable, subjective descriptions does not establish that rewriting a page will increase funding. Nor does subjectivity mean abandoning specifications. Personal explanations can accompany clear technical evidence.

Platforms could test disclosure formats that distinguish promotional assistance from AI embedded in the product itself. Any experiment should measure comprehension and informed participation, alongside funding outcomes. A label that increases pledges while confusing backers would offer a poor measure of transparency.

What Investors Can Learn From Crowdfunding Trust

The research concerns reward-based crowdfunding. Backers are not necessarily buying ownership or seeking investment returns. Securities.io’s guide to equity crowdfunding, updated in 2026, explains the separate model in which companies sell securities. The Kickstarter estimates should not be transferred directly to those offerings.

Nevertheless, the study provides a useful question for assessing AI businesses: does a product reduce the customer’s total burden, including verification? Faster content creation can generate additional review work if users must check accuracy, rights, or the distinction between concepts and finished products.

That creates a potential market for tools combining generation with documented workflows and provenance. Provenance records describe how content was created or edited. They do not independently establish that the content is true, that every legal issue is resolved, or that a company can fulfill its promises.

Adobe Connects Generative AI With Content Provenance

Adobe offers publicly traded exposure to this intersection through its creative software, Firefly tools, and Content Credentials initiatives. Its relevance comes from serving content production and transparency within the same broader software ecosystem.

In März 2026, Adobe (ADBE ) announced a strategic partnership with NVIDIA covering next-generation Firefly models and creative and marketing workflows. The announcement describes development plans, rather than proof that every proposed capability has been delivered.

Adobe has also introduced Content Authenticity for Enterprise, including services for integrating provenance into creative and publishing workflows. Its Firefly documentation describes Content Credentials as tamper-evident metadata that can provide context about creation and editing.

The investment thesis is that professional customers may value tools that help document production as well as accelerate it. However, the Kickstarter study does not demonstrate demand for Adobe products. Investors still need evidence that these capabilities support customer retention, paid adoption, and profitable growth amid competition.

Latest Adobe (ADBE) News and Developments

AI Productivity Still Depends on Customer Confidence

The study remains observational. Matching improves comparisons on measured characteristics, but unobserved project differences could influence results. Absence of a disclosure section also does not prove absence of AI use. The researchers did not directly observe the backer perceptions proposed to explain the funding gap.

Within those limits, the research highlights a practical challenge: AI can make communication more impressive without making execution easier to evaluate. Creators and software providers have an opportunity to close that gap through clearer explanations and verifiable evidence. As production costs fall, demonstrating responsibility may become an increasingly valuable part of the product.

References:

¹ Singhal, S., Kim, Y., Kang, S., & Yoon, V. (2026). Will your AI turn backers away? Evidence from Kickstarter’s AI-use disclosure policy. Decision Support Systems, 211, Article 114773. https://doi.org/10.1016/j.dss.2026.114773