Independent reviews
Image generators
Quality, control, provenance, commercial terms, and whose risk is being externalized.
Compare the field →
Synthetic identity, examined
When a person can be generated, what remains real?
We test the tools, trace the risks, and ask harder questions about images, voices, identity, consent, and the generated self.
Published 2025-06-01 · Updated 2026-08-28 · About
Independent reviewsSources linked, limits named
Consent-first analysisCapability never erases responsibility
No hype rankingsWe score tradeoffs, not spectacle
$1 to beginLow friction, real member access
The field guide
Generated media is not one problem.
It is a cluster of creative breakthroughs, labor disputes, security failures, accessibility gains, consent questions, and old human incentives wearing very new machinery.
Consent required
Voice cloning
Verified-service depth plus open-weight coverage, risk tiers, and safer workflows for accessibility and production.
Read the guide →The harder question
The generated self
Likeness, continuity, identity, proof, memory, and who gets to speak in your shape.
Enter the ethics archive →A practical standard
Consent is not a checkbox someone else can tick.
- Identity is a source, not raw material.
A face or voice being accessible does not make it ethically available. - Disclosure belongs near the media.
Not buried in terms, metadata, or a defense issued after harm. - Verification must survive distribution.
Provenance metadata helps; resilient watermarking and social context matter too.
Membership
Serious work needs readers, not a velvet rope.
Explorer costs $1 a month. It unlocks full reviews and makes the next piece possible. Higher tiers fund deeper comparisons, member discussion, and manual research.
Deeper research — plainly described
We score trade-offs, then show our work.
Public pages give you the comparison table, verdicts, and date-stamped sources. Member pages add the parts that don't fit in a teaser: tested prompts, failure cases, deletion checks, and a versioned decision matrix you can reuse.
Image pipeline detail
Prompt-collapse notes, privacy/deletion checks, and the full decision matrix with tier recommendations — dated and sourced, not a lab prop.
Voice-workflow detail
Open-weight vs. verified-service trade-offs, revocation language, disclosure placement, and risk tiers for accessibility and production use.
Corrections & deltas
When a model, plan, or policy moves the needle, we log it in the public update feed. Outdated claims are easy to flag: editor@persongenerated.com.
Scope note
This site is deliberately narrow: image generators, one verified voice path plus open-weight coverage, and the ethics archive. No video, no full persona/agent systems. If you need broad “AI people” tooling, we say so plainly — and link to where to look next.