Botanical Fraud in the Age of AI: A Technical Guide to Identifying Synthetic Plant Imagery

Direct Answer and Scope

Botanical fraud listings succeed when a synthetic or stolen image is treated as biological evidence. A technical buyer can reduce risk by running a simple forensic protocol: verify pigment biology, verify taxonomic legitimacy using authoritative databases, then test the image for generative artifacts and camera provenance.

This article is a long form forensic analysis for plant shoppers and educators. It is written in an investigative and educational style. It focuses on two high volume fraud themes, blue strawberries and rainbow or blue roses, then generalizes the method to other synthetic plant imagery.

Experiment at a Glance

Objective	Method	Evidence to collect	What a real listing looks like	What a fraudulent listing looks like
Test biological plausibility	Pigment pathway check	Anthocyanin class, genus constraints	Red family pigments in strawberry, documented rose pigments	Delphinidin based blue claims in strawberry and rose seeds
Test taxonomic legitimacy	Database verification	Accepted name, synonyms, distribution	Names resolve in IPNI and POWO	Names do not resolve, or names used incorrectly
Test image authenticity	Generative artifact scan	Venation geometry, highlight physics, aliasing	Irregular detail, consistent light	Edge melting, cloned patterns, impossible specular highlights
Test provenance	Metadata and reverse search	EXIF fields, duplicates, timestamps	Seller provides original series	Same image appears across unrelated sellers

Definitions Used in This Guide

This guide uses a few terms in a precise way.

Synthetic plant imagery means an image produced by a generative model, or an image significantly altered so it depicts a phenotype that did not exist in the original scene.

Botanical fraud means a commercial attempt to sell seeds, plants, or cuttings using synthetic or misleading evidence about what the buyer will receive.

Phenotype claim means what the listing says the plant will look like. For example, blue strawberry fruit or a rose that blooms in multiple colors in one flower.

Provenance means the chain of custody of media. It is the story of who captured it, when, where, and whether it can be independently validated.

Section 1. The Biology of Pigmentation

Anthocyanins, anthocyanidins, and what color means in plant tissues

Anthocyanins are water soluble pigments stored in the vacuole of plant cells. Anthocyanins are glycosylated forms of anthocyanidins. The sugar groups increase stability and solubility.

Anthocyanin color is not a simple pigment equals color equation. Color depends on:

1. The specific anthocyanidin core, such as pelargonidin, cyanidin, or delphinidin
2. The pH of the vacuole
3. Co pigments, such as flavonols, that shift hue through molecular stacking
4. Metal ion complexing in some species and tissues
5. Tissue structure, including cell shape and surface scattering

Even with these variables, there are strong genetic constraints at the genus level. Those constraints are what make certain viral colors unrealistic.

The anthocyanin pathway at a technical but shopper useful level

The pathway relevant to red and blue anthocyanins is part of the phenylpropanoid and flavonoid network. A simplified sequence is:

1. Phenylalanine enters phenylpropanoid metabolism
2. Chalcone synthase and chalcone isomerase produce flavanones
3. Hydroxylases and reductases shift the B ring hydroxyl pattern
4. Anthocyanidin synthase forms the anthocyanidin core
5. Glycosyltransferases stabilize pigment into anthocyanins
6. Transport into the vacuole accumulates color

The most important fork for blue claims is the hydroxylation pattern on the B ring. Delphinidin requires the correct hydroxylation pattern and enzyme capability in the tissue.

Why delphinidin based blue is genetically implausible in Fragaria fruit

The cultivated strawberry is Fragaria × ananassa. Strawberry fruit color is dominated by pelargonidin based anthocyanins, with cyanidin based pigments present in some varieties.

A stable blue fruit claim requires either:

1. Delphinidin based anthocyanins accumulating at high levels in the fruit tissue
2. Or a structural color mechanism that produces blue appearance without blue pigment

In strawberry fruit, the delphinidin route is not a standard outcome. Strawberry fruit chemistry and genetics are aligned with red family pigments and acidic vacuoles. This strongly constrains hue.

A common technical summary is:

1. Strawberry fruit anthocyanin profiles are typically pelargonidin heavy
2. Delphinidin accumulation in fruit is not documented as a stable consumer phenotype
3. Vacuole pH in fruit is generally acidic, which pushes anthocyanins toward red

For a buyer, the practical interpretation is direct.

If a listing claims seed grown strawberries that reliably produce blue fruit, the claim conflicts with known pigment pathway constraints in Fragaria fruit.

Why true blue delphinidin roses are also not a credible seed claim

Roses are complex and have many pigments. Many modern roses show red, pink, purple, and orange. Some appear lavender. The viral blue rose claim is different. It usually shows a saturated cobalt blue that reads as primary blue.

A true blue rose appearance requires pigment chemistry and tissue conditions that roses do not naturally express in the same way that naturally blue flowers do. There have been high profile attempts to engineer blue leaning roses, but those are controlled, proprietary products, not random seed packets in online marketplaces.

A seed listing that claims blue roses with saturated true blue petals is not credible unless it provides:

1. A traceable breeder
2. A cultivar name with documentation
3. Independent third party confirmation
4. A consistent series of photos and provenance

Most fraudulent listings provide none of these.

Why rainbow seeds will not produce advertised phenotypes

Rainbow seed listings often imply one of two things.

1. A single plant produces multicolor fruit or flowers at once
2. Or a mixed seed packet will produce different colors and the image shows a bouquet result

The first claim is usually fraudulent when the image shows a single strawberry that is blue or a single rose with multiple colors in a perfect spectrum pattern. Developmental genetics can create patterns, but the viral patterns often violate how pigmentation domains form.

The second claim can be misleading even when it is partly true. Mixed seed packets can produce different colors across different plants. They do not guarantee that each individual plant will match a staged marketing image.

If a listing shows a single plant or bloom with a perfect rainbow distribution, treat it as a high risk claim and demand very strong proof.

Section 2. Anatomy of an AI Image

This section focuses on forensic signals that appear when diffusion models generate plant imagery. The goal is not to name a tool. The goal is to identify failure patterns that humans can reliably spot.

Diffusion model aliasing in plant edges and fine textures

Aliasing is a sampling artifact. In synthetic images, aliasing often appears as:

1. Stair step edges along serrated leaves
2. Repeating pixel patterns in fine hairs and trichomes
3. Moire like banding in petal gradients
4. Small detail that looks sharp but breaks into blocky fragments when zoomed

Real camera aliasing exists too, but it tends to be consistent with sensor and lens behavior and it usually appears with predictable noise.

In diffusion outputs, aliasing can be inconsistent across the image. One leaf edge may show stair stepping while the adjacent edge looks painted.

Inconsistent specular highlights that violate surface physics

Specular highlights are reflections of a light source on a glossy surface. Leaves and fruit can be glossy, so highlights are normal. The forensic question is whether the highlight geometry matches the light field.

A physically plausible scene tends to have:

1. One dominant highlight direction outdoors
2. Highlights that move smoothly with surface curvature
3. Similar highlight direction across objects in the same plane

Diffusion outputs often show:

1. Highlights on multiple sides of the same berry
2. Highlights on leaves that do not match highlights on the fruit
3. Highlight shapes that are identical on multiple leaves, suggesting cloning
4. Bright rim light with no source and no corresponding shadow behavior

These failures are more visible when you compare multiple leaves in the same image.

Geometrically impossible leaf venation

Leaf venation is an ideal forensic target because it follows biological architecture and is hard to fake consistently.

A real leaf venation system shows:

1. A midrib that connects to the petiole
2. Secondary veins that branch in consistent angles
3. Smaller veins that fill the lamina in a nested pattern

Synthetic images often show venation that is detailed but wrong.

Common failure modes include:

1. Secondary veins that start without connecting to the midrib
2. Veins that loop and reconnect in ways that do not match plant anatomy
3. Veins that terminate abruptly in the middle of the leaf
4. Mirror symmetric venation across multiple leaves, suggesting generative copying
5. Petiole attachment points that do not match the start of the midrib

When a strawberry listing shows perfect blue fruit plus leaves with impossible venation, you have two independent lines of evidence for fraud.

Diffusion edge fusion and object melting in overlaps

Diffusion models can struggle at boundaries where objects overlap.

Look for:

1. Leaf margins that blend into adjacent leaves
2. Petals that fuse at overlap seams
3. Stems that split then rejoin
4. Calyx structures that deform into the fruit surface

These are structural failures, not just beauty filters.

A practical scoring protocol for shoppers

Assign 0 to 2 points per category.

0 means absent
1 means mild
2 means obvious

Categories:

1. Aliasing and edge stair stepping
2. Specular highlight inconsistency
3. Venation geometry errors
4. Edge fusion or melting
5. Repeated cloning patterns

A total of 5 or more should be treated as a strong synthetic signal unless there is excellent provenance and independent confirmation.

Section 3. The Economics of Seed Fraud

Synthetic imagery is not the whole fraud. The business model matters.

The drop ship seed supply chain

A typical drop ship seed scam supply chain is:

1. A storefront is created on a marketplace platform
2. Synthetic images are generated or stolen and attached to listings
3. Low priced seed packets are offered with high quantity claims
4. Orders are forwarded to a fulfillment source that ships generic seed or low value seed
5. When buyers complain, the seller delays by citing growth time, shipping, or customs

This model scales because it has low overhead.

Why rainbow seeds fail at the phenotype level

Rainbow seed listings often fail because:

1. Seeds are genetically variable
2. Many advertised traits require controlled breeding and clonal propagation
3. The listing shows a phenotype that is not stable or not biologically plausible

In strawberries, named varieties are often propagated by runners because seed does not reliably reproduce the same fruit qualities. In roses, seeds rarely reproduce the exact parent bloom and take time to reach maturity. A seed packet cannot guarantee a specific staged photo result.

When a listing promises guaranteed rainbow results from seed, the promise is a technical claim. It is a claim about inheritance. In most cases, it is not defensible.

Why this remains profitable

The economics are simple.

1. Each transaction is small, so many buyers do not fight.
2. Growth time delays disputes.
3. Platforms can be slow to remove sellers.
4. New accounts can be created quickly.

Forensic awareness reduces the conversion rate. That is one of the few direct pressures consumers can apply.

Section 4. Database Verification Protocols

This section provides a protocol for verifying plant names using authoritative databases. The goal is to confirm that a botanical name exists and is used correctly.

Important limitation: IPNI and POWO focus on plant names and taxonomy. They are not complete cultivar registries for every horticultural trade name. However, they are still powerful tools for detecting nonsense names and mismatched genus claims.

Step by step protocol using IPNI

The International Plant Names Index is a nomenclatural database. It is useful for verifying whether a scientific name has been published and whether the author citation and publication details are real.

Procedure:

1. Identify the claimed scientific name from the listing
2. Go to IPNI and search the genus and species
3. Confirm the name exists and note the author and publication details
4. Check for synonyms and accepted names
5. Confirm the genus and species spelling matches accepted usage

Interpretation:

1. If the name does not exist, the listing may be using invented taxonomy.
2. If the name exists but the listing uses it incorrectly, the seller may lack botanical competence.
3. If the seller will not provide a scientific name at all, the listing is weak.

Step by step protocol using Kew Plants of the World Online

Plants of the World Online is maintained by Kew and provides accepted names, synonyms, and distribution information.

Procedure:

1. Search the genus and species
2. Confirm the accepted name
3. Review the distribution information
4. Compare distribution to the seller origin claim
5. Review synonyms that may explain alternate naming

Interpretation:

1. If the seller claims a species from a region where POWO indicates it is not native or not present, ask for documentation.
2. If the listing claims a plant that is actually a different genus, treat it as an identification failure.

How to use database verification in real shopping decisions

Use the databases as filters.

1. Filter out invented names and obvious misuse
2. Identify whether the species claim is coherent
3. Then proceed to provenance and image forensics

Do not treat database presence as proof of cultivar legitimacy. Treat it as a minimum standard for basic scientific naming.

Section 5. Real Rare Cultivars vs AI Fakes

Fraud listings often borrow the language of real rarities. A buyer should know the difference between unusual but real plants and impossible plants.

Example 1. Pineberry versus blue strawberry listings

Pineberry is a real strawberry type often described as white to pale pink fruit with red achenes and a pineapple like aroma. It fits known strawberry pigment biology. It is a low anthocyanin phenotype rather than a new blue pigment phenotype.

A pineberry listing should show:

1. White fruit with red achenes
2. Normal strawberry leaves and flowers
3. A realistic description of taste and yield
4. A credible seller who ships runners or plants more often than seed

A blue strawberry listing often shows:

1. Saturated cyan fruit with uniform coloration
2. Perfect symmetry and glossy surfaces
3. Vague claims about rare origin
4. Seed packets as the primary product

These two categories are not comparable. One is a plausible pigment reduction phenotype. The other is a pigment class claim that conflicts with genus constraints.

Example 2. Hylocereus and other real exotic fruits versus AI composites

Hylocereus, often called dragon fruit, is a real cactus genus with striking fruit. Real dragon fruit photos can look unusual to new buyers, which makes it a useful comparison category.

A real Hylocereus listing should show:

1. The cactus growth habit and climbing form
2. Large night blooming flowers in some evidence sets
3. Fruit attached to stems in ways consistent with cactus anatomy

AI fakes often show:

1. Fruit attached to woody trees that do not match cactus anatomy
2. Leaves that resemble unrelated species
3. Impossible combinations of flower and fruit at the same time

The lesson is that real rare plants still obey anatomy.

Comparative checklist: Real rarity indicators

Real rarity listings tend to include:

1. A stable scientific name that resolves in databases
2. A propagation method that matches the plant, such as cuttings for some plants, runners for strawberries
3. Realistic pricing for scarcity and shipping
4. Media that shows context and time series growth

AI fake listings tend to include:

1. Vague naming and invented cultivar claims
2. Seed only offers for traits that are typically clonal
3. Extremely low price with huge seed counts
4. Synthetic image artifacts and reused images across sellers

Section 6. Consumer Protection Framework

This section outlines a practical dispute workflow. It is not legal advice. It is a consumer documentation guide.

Evidence you should collect before filing any dispute

Collect evidence before you message the seller too much. You want clean documentation.

1. Screenshots of the listing, including title, price, and claimed phenotype
2. Screenshots of the photos and any description of rarity or guarantees
3. Order confirmation and payment records
4. Shipping label photos if the product arrives
5. Photos of received seeds, packaging, and any labels
6. Your reverse image search results, including URLs
7. Your database verification notes from IPNI and POWO

Filing with the FTC

The Federal Trade Commission collects fraud reports and can use them for enforcement patterns.

Procedure:

1. Go to reportfraud.ftc.gov
2. File a report using the category that best matches the transaction
3. Include the seller name, platform, URLs, and a clear description of what was misrepresented
4. Upload or retain evidence as instructed

A strong report is specific and evidence based.

Filing with the BBB

The Better Business Bureau can be useful for tracking patterns and pressuring some businesses to respond.

Procedure:

1. Locate the business profile if it exists
2. File a complaint with clear claim language
3. Include transaction dates and evidence
4. State the resolution you want, such as refund

BBB coverage depends on whether the business is identifiable. Many marketplace sellers are not.

Filing platform disputes

Platform tools are often the fastest way to get money back.

Etsy

1. Use the order help and case system
2. Select item not as described when the phenotype claim is impossible or misrepresented
3. Provide screenshots and reverse search evidence

eBay

1. Use eBay buyer protection workflows
2. File an item not as described claim
3. Provide clear side by side claim and evidence

Facebook Marketplace

1. Report the listing for scam behavior
2. Report the seller profile if repeated
3. Use platform payment dispute tools when available
4. Document conversations because many interactions happen in messaging

Payment provider disputes

If the platform dispute fails, payment disputes can be effective.

1. Credit card chargebacks require clear documentation and fast action
2. PayPal disputes use the resolution center and benefit from strong evidence

Post dispute steps

If you receive seeds of unknown origin, do not plant them. Keep them sealed and consult local guidance if you suspect they are illegal imports or mystery seeds.

Frequently Asked Questions

Q1: What is botanical fraud in the age of AI?
Botanical fraud is the sale of seeds or plants using synthetic, stolen, or misleading imagery to claim a phenotype the buyer will not receive.

Q2: What makes an image synthetic instead of edited?
Synthetic images are generated from scratch by a model. Edited images start from a real photo but can still be fraudulent if the edit creates a false phenotype claim.

Q3: Why are delphinidin based blue strawberries not credible?
Because cultivated strawberry fruit pigment profiles are dominated by red family anthocyanins and do not support a stable consumer blue phenotype based on delphinidin accumulation.

Q4: Why are blue rose seed listings not credible?
Because true saturated blue petals are not a normal rose phenotype, and legitimate blue leaning roses are not sold as random seed packets with no breeder documentation.

Q5: What does anthocyanin mean in practical shopping terms?
It means color is constrained by chemistry. If the pigment pathway cannot make the claimed pigment, the listing is likely false.

Q6: What is the single best botanical test for a viral plant photo?
Ask which pigment class is being claimed and whether that genus is known to produce it in that tissue.

Q7: What are the most reliable diffusion model artifacts in plant images?
Edge fusion at overlaps, inconsistent specular highlights, and venation geometry that fails anatomical rules.

Q8: What does aliasing look like in fake leaf edges?
It looks like stair steps or blocky jagged edges that do not match a camera focus plane.

Q9: What are impossible specular highlights?
Highlights that appear on multiple sides of the same object or do not match the direction of shadows and other reflections.

Q10: What is the fastest way to detect a stolen plant image?
Reverse image searching the full image and multiple crops.

Q11: Is missing EXIF proof of fraud?
No. Many platforms remove EXIF. It is a weak negative signal, not proof.

Q12: What EXIF fields are most useful when present?
Date time original, camera model, and software field.

Q13: Can EXIF be faked?
Yes. Treat it as one line of evidence, not the final answer.

Q14: Why do scam listings prefer seeds?
Because seeds are cheap to ship, hard to verify quickly, and disputes often arrive late.

Q15: Why will rainbow seeds never match the photo?
Because the photo usually depicts a staged or synthetic phenotype that is not genetically stable, and seed inheritance rarely reproduces exact ornamental outcomes.

Q16: How do I use IPNI to verify a plant name?
Search the scientific name, confirm it exists, and note author and publication details. Names that do not resolve are suspect.

Q17: How do I use POWO to verify a plant claim?
Confirm the accepted name and compare distribution information to the listing origin claim.

Q18: If a name exists in IPNI, does that make the seller honest?
No. It only shows the name is real. Fraud can still occur through mislabeling and synthetic images.

Q19: What is a real example of a rare but plausible strawberry phenotype?
Pineberry, which is white to pale pink, fits known strawberry pigment constraints.

Q20: What is a real example of an unusual fruit that is often faked?
Dragon fruit in Hylocereus, which is real but sometimes shown in AI composites attached to incorrect plant bodies.

Q21: What is a drop ship seed scam?
A storefront that sells seeds without controlling inventory, using synthetic marketing images while fulfillment ships generic seed.

Q22: What evidence should I collect before filing disputes?
Screenshots, reverse search results, seed packaging photos, and any database verification notes.

Q23: How do I file with the FTC?
Use reportfraud.ftc.gov and include seller, platform, URLs, and clear misrepresentation details.

Q24: How do I file with the BBB?
File a complaint if the seller is identifiable, with transaction details and desired resolution.

Q25: How do I dispute on Etsy or eBay?
Use item not as described and attach evidence. Be precise about the phenotype claim.

Q26: What should I do with mystery seeds?
Do not plant them. Keep packaging and consult local agriculture or extension guidance.

Q27: Can a seller prove legitimacy without EXIF?
Yes. A time series, one take video, and consistent original context images can establish strong provenance.

Q28: What is the best single request to make to a seller?
Ask for a one take video that begins with a wide shot of the plant and moves to the claimed phenotype with no cuts.

Q29: What does it mean when the same photo appears on many unrelated sites?
It indicates the photo is not unique evidence and may be stolen or synthetic.

Q30: What should I do if I already planted suspicious seeds?
Document the process, isolate the plants from other collections when possible, and report platform fraud if the claim was misrepresented.

References

1. USDA NRCS. USDA PLANTS Database. https://plants.usda.gov
2. Missouri Botanical Garden. Plant science resources and Plant Finder. https://www.missouribotanicalgarden.org
3. Royal Horticultural Society. Plant and gardening resources. https://www.rhs.org.uk
4. Federal Trade Commission. Fraud reporting portal. https://reportfraud.ftc.gov
5. Federal Trade Commission. Consumer advice on scams. https://consumer.ftc.gov
6. USDA APHIS. Import and plant health guidance. https://www.aphis.usda.gov
7. International Plant Names Index. Plant name verification. https://www.ipni.org
8. Kew. Plants of the World Online. https://powo.science.kew.org
9. International Plant Protection Convention. Phytosanitary framework overview. https://www.ippc.int
10. NCBI PubChem. Anthocyanidin compounds and properties. https://pubchem.ncbi.nlm.nih.gov
11. Better Business Bureau. Scam Tracker and complaint resources. https://www.bbb.org
12. Wikimedia Commons. EXIF field explanations. https://commons.wikimedia.org/wiki/Commons:Exif

---

Technical Disclaimer: This article is educational and does not provide legal advice. Regulations, platform policies, and import requirements can change. Confirm current USDA APHIS guidance and your state agriculture requirements before importing or planting seeds. Tierney Family Farms does not endorse third party sellers or listings.