Why Gemini Restricts Person Image Generation: Policy, Account Conditions, or Quota?

Gemini can generate images of people and edit photos users upload, and all users generate images with Nano Banana 2. Google even lists trying new hairstyles or placing yourself in different scenes as an intended use. So the claim that Gemini cannot draw people is inaccurate. When a request is refused, the first step is identifying the cause, because each category requires a different response: a policy block, an account condition, or a quota or service problem.

Policy blocks apply when a request involves unconsented use of someone else's likeness, private images, minors, sexual or graphic violent content, or impersonation for deception. There is no compliant way around these. Account conditions include not being logged in, being under 18 (who cannot edit images), an unsupported country or language, or restrictions on work or school accounts; these are resolved by meeting the conditions. Quota or service issues include exhausting the daily Nano Banana 2 allowance or a temporary outage, which have nothing to do with whether people appear in the image. Google does not publish exact daily image counts; free usage is compute-based, refreshes every 5 hours, and has a weekly cap.

The Generative AI Prohibited Use Policy bans using someone's likeness without consent, creating non-consensual intimate imagery, and impersonating a person, living or deceased, to deceive. It also prohibits circumventing safety protections such as prompt injection. The Gemini app policy guidelines add concerns around child safety, dangerous activities, graphic violence, harmful factual errors, harassment, and sexual content, while noting that educational, documentary, artistic, or scientific context is considered. The key factors are consent, privacy, and whether the image could mislead viewers. Editing your own photo is explicitly encouraged; sexualized or dangerous content involving minors is never allowed.

As of October 5, 2026, Google does not publish a list of banned celebrities or public figures for the Gemini app, nor refusal rates. Users widely report refusals when naming public figures, but this is observed behavior rather than written policy. Google's last explicit statement, from August 28, 2024, said it does not support generating realistic identifiable individuals, and that text describes an older model. Removing a name and substituting a detailed appearance description does not change the nature of the request, since you are still depicting that real person.

Image generation in the Gemini app requires signing in with a Google account. Personal accounts must be at least 13 (or the age set by local law) to generate images, and at least 18 to edit images. Image generation is available only in supported countries and languages; as of October 5, 2026, mainland China is listed as Workspace-only, so personal accounts are not eligible, while Hong Kong and Taiwan appear in the list. Work or school accounts may face different terms set by administrators. The Gemini API and Google AI Studio have their own region restrictions, which are eligibility conditions with no technical workaround.

Gemini app usage is compute-based and refreshes every 5 hours until a weekly cap is reached. Standard access applies without a subscription; Google AI Plus provides 2x, Google AI Pro 4x, and Google AI Ultra 5x or 20x. All users default to Nano Banana 2, while Nano Banana Pro is available to Google AI subscribers through the Redo with Pro option. When the Nano Banana 2 daily allowance runs out, Redo with Pro is also unavailable. If a simple request without people also fails, the cause is likely a service issue rather than the people rules.

Editing your own photos is the recommended approach, including changing hairstyles, scenes, or backgrounds, which requires login and age 18 or older. For photos of others, obtain their consent first; parental consent is needed for children's photos. The avatar feature lets you record your face and voice to generate images of yourself consistently. Original fictional characters are permitted if described by age range, appearance, clothing, action, scene, lighting, and purpose. Clear, specific prompts reduce misinterpretation. If compliant requests are repeatedly refused, switch to your own photo or an original character rather than rewording prompts to probe the filter, since circumventing safety protections is itself a violation.

In February 2024, Google paused Gemini's person image generation after inaccurate and offensive historical images, announcing the pause on February 23. Person generation returned on August 28, 2024 with Imagen 3, for Gemini Advanced, Business, and Enterprise users in English. Nano Banana and Nano Banana Pro launched in August and November 2025, shifting the app toward editing use cases. On February 26, 2026, Google released Nano Banana 2, replacing Nano Banana Pro in most Gemini app models; its announcement did not introduce new person restrictions and mentioned SynthID watermarks and C2PA content credentials. Users reporting sudden refusals may be seeing the model change, and some developer reports in March 2026 described new safety filtering, though Google has not published a change log for the filters.

The Gemini image models have no person-generation toggle. The personGeneration parameter belonged to Imagen, which is now retired on the Gemini API, with gemini-3.1-flash-image as the replacement. Google Search grounding for gemini-3.1-flash-image does not currently support real people's images from web search. API requests are subject to the same policy as the app. Blocked requests are diagnosed through promptFeedback.blockReason (such as SAFETY, PROHIBITED_CONTENT, BLOCKLIST, IMAGE_SAFETY, or OTHER) and finishReason (such as IMAGE_SAFETY, IMAGE_PROHIBITED_CONTENT, SPII, or IMAGE_OTHER). Adjustable safety settings do not include core child safety protections, and setting thresholds to BLOCK_NONE is unlikely to lift the other categories. Google has acknowledged that OTHER errors can be false positives and plans to redesign the system, so legitimate blocked requests should be reported to Google with the response fields preserved.

In the Gemini app, confirmed violations trigger in-product or email notices, and repeated violations can restrict generative AI features or affect the whole Google account. Users can appeal through links in the notice or email. For the Gemini API and AI Studio, Google may contact users by email, temporarily adjust rate limits or route to other models, suspend access, and eventually disable access to Gemini API and other Google services, with appeal links included in suspension and closure notices. Prompts, context, and outputs may be retained for 55 days for abuse detection.

Q: Can Gemini's person image restrictions be lifted? A: Policy restrictions cannot be lifted, and there is no switch to turn them off. Account conditions and quotas can be resolved by signing in, meeting the age requirement, using a supported region, or subscribing to a Google AI plan. Q: Can Gemini generate images from my own photos? A: Yes, if you are signed in and 18 or older. Q: Can Gemini generate historical figures? A: No specific rule exists, but policies on impersonation and harmful factual errors apply, and historical figure requests are likely to be refused. Q: Is Gemini image generation free? A: The Gemini app offers standard access without a subscription at 1K, with higher limits and 2K downloads for subscribers; API image output has no free tier. Q: How many images can Google AI Pro generate per day? A: Google does not publish an exact number; limits are compute-based, refresh every 5 hours, and have a weekly cap.