GEO ethics in 2026: no framework, growing stakes.

The ethical imperative iPullRank named

iPullRank’s Michael King has done the most rigorous published thinking on the GEO ethics question. The frame: “engineer relevance responsibly, or allow the machines to engineer our reality for us.”

The choice is not theoretical. AI systems “perform truth rather than presenting it” — they generate single authoritative-sounding answers without the disclaimers and confidence-interval signaling that academic research carries. Users receive AI outputs with disproportionate trust. Manipulation of those outputs has disproportionate consequences.

The structural problem: no industry body has adopted a code of ethics for GEO. No certification, no standards, no enforcement mechanism. Each practitioner makes their own decisions about where the line sits. The most credible operators have published their own frameworks; the least credible have not.

“The invisible algorithm's most visible impact may be whether we choose to engineer relevance responsibly or allow the machines to engineer our reality for us. — iPullRank”

TigerTracks’ line

TigerTracks proposed the cleanest operational distinction in the published literature: optimization helps the engine perform its duty to the user; manipulation hijacks the model’s decision logic through noise.

The test is functional, not procedural. Three diagnostic questions for any GEO tactic:

1. Is the underlying claim true? If the optimization makes a true claim more findable, it’s helping the engine. If it makes a false claim more findable, it’s manipulating it.
2. Would a knowledgeable human reviewer agree your brand belongs in the answer? If yes, the tactic is structural advocacy for an honest position. If no, the tactic is engineering an outcome the evidence doesn’t support.
3. Does the tactic depend on the AI engine not noticing? If yes, it’s manipulation by definition. If the same tactic survives full transparency to the platform, it’s optimization.

Tactics that pass all three tests: structured content, schema markup, accurate brand entity definition, earned-media outreach, original research publishing. Tactics that fail any one test: synthetic consensus generation, fake reviews, hidden prompt injection, fabricated credentials.

The bias dimension

UNESCO’s ongoing work on AI ethics has surfaced a less-discussed GEO concern: AI systems amplify the biases embedded in their training data. Gender, racial, cultural, and ideological biases that exist in the underlying corpus get propagated into AI recommendations.

Stanford GSB research on AI political bias found that both Republicans and Democrats perceived left-leaning bias in LLMs’ discussion of contentious topics. The directional finding is less important than the structural one: AI engines are not neutral arbiters. They reflect the perspectives of the corpora they were trained on and the humans who reinforced them.

For brands operating in politically or culturally sensitive categories (healthcare, education, financial services, news), the implication is operational: even unimpeachable optimization tactics can produce outputs that some user segments perceive as biased. The defensive read: brands need to monitor not just whether they’re cited but in what framing.

The trust paradox

Exploding Topics’ 2025 consumer-trust research surfaced the paradox that defines the user-side ethics question:

• 82% of users are skeptical of AI results.
• Only 8% always verify sources.
• Forbes found AI search results are more trusted than ads; consumers describe AI as “less cluttered” than search.
• 37% of consumers now start searches with AI instead of Google (Search Engine Land).

The pattern: users distrust AI in principle and depend on it in practice. The skepticism does not translate to verification behavior. The result is the worst-case dynamic for GEO ethics — a user base that knows AI can be wrong but acts as if it were right, and a brand-side incentive structure that rewards the brands whose optimization is most aggressive whether or not the underlying claims are accurate.

What ethical operators publish

In the absence of an industry framework, the most credible operators have published their own ethical commitments. The patterns that emerge across them:

1. Disclose methodology. If a vendor cannot explain in plain language what they do for clients, the methodology is probably either trivial or dishonest. The credible operators publish their playbooks.
2. Cite primary sources. Claims about AI engine behavior should be backed by named studies, not asserted. The credible operators link to their evidence.
3. Refuse synthetic-volume tactics. Mass AI-generated content, fake reviews, synthetic personas. The credible operators publicly reject these.
4. Document client commitments. What you will and will not do on a client’s behalf, in writing, before the engagement starts.
5. Acknowledge uncertainty. AI engines are evolving black boxes. Vendors who claim certainty about future behavior are either selling something or not paying attention.

Aleyda Solis, who runs Orainti and the SEOFOMO newsletter (35K+ subscribers), has been among the most consistent published voices on the ethics question. Her central warning: “If you treat AI search solely as a performance channel — expecting traffic and revenue from every inclusion in AI answers — you’ll set yourself up for disappointment.” Treating GEO purely as performance optimization, not as relationship-building with the information ecosystem, is the path that leads operators across the ethical line.

The buyer’s diligence framework

Until an industry body publishes a code of ethics, buyers are the de facto enforcement mechanism. The questions a serious buyer should ask any GEO vendor:

1. Walk me through every type of action you take on a client’s behalf. Specific, not generic. Get the list in writing.
2. Which of these actions involve creating content, citations, or signals that originate from sources other than the client? If the answer is “none,” the engagement is probably purely advisory. If the answer involves third parties, dig in.
3. Do you publish synthetic personas, AI-generated reviews, or coordinated comment campaigns? The honest answer should be “no.” If it isn’t, walk away.
4. How do you measure success? If the answer involves “ranking position,” the vendor either misunderstands AI engines or is misleading you.
5. What happens to the brand’s reputation if a platform flags any of your tactics? The vendor should have thought about this. If they haven’t, the buyer is carrying the risk alone.

The brands that emerge from the next 24 months with strong AI visibility and intact reputations will be the ones whose vendors could answer those questions clean.

Every statistic in this piece follows Visigent's research standards: named sources, dated figures, corrections on request to help@visigent.com.