AI Safety
AI Safety Policy
AgentForge is built for responsible commercial use — product photography, fashion mockups and catalogue creation. This page explains what content we block, what guardrails sit between you and the model, your responsibilities as a user, and how to report misuse.
Last updated
May 2026 · This policy may evolve as we add new agents and capabilities.
What AgentForge blocks
✕Nudity, sexual or adult content
We block generation of nude, semi-nude, sexualised or adult-themed content. Any prompt or image that attempts to bypass this guardrail is rejected and may trigger an account review.
✕Minors in unsafe scenarios
We do not permit generations that depict minors in mature, sexualised, dangerous, or otherwise unsafe contexts. Kidswear product mockups are fully supported in family-safe settings — that is a different category.
✕Celebrities, public figures and politicians
We do not allow impersonation of named celebrities, public figures or political leaders. The AI flags both prompts ("in the style of [name]") and uploaded portrait images of recognisable figures.
✕Deepfake or non-consensual identity use
Uploading someone's face/identity without their consent is strictly prohibited. UGC Forge requires confirmation that you have the creator's permission before generation begins.
✕Hate, harassment, violence and weapons
Content that promotes hatred against a community, harasses individuals, glorifies violence, or depicts weapons in harmful contexts is blocked.
✕Illegal goods or activities
Mockups for illegal substances, counterfeit goods, fake brand impersonation ("make this look like the real Nike logo"), or fraudulent product claims are not supported.
✕Misinformation or impersonation of real entities
We don't generate visuals that misrepresent real news events, fake government communications, or impersonate trusted institutions.
Our guardrails
🛡️Prompt-level filtering
Every prompt is screened by an upstream safety classifier before it reaches the generation model. Unsafe instructions never enter the pipeline.
🛡️Image-level checks
Uploaded images are scanned for nudity, minor faces in unsafe contexts, and known public-figure faces. Matches are blocked at upload, not after generation.
🛡️Post-generation moderation
Outputs go through a final NSFW + identity check. If the model produces unsafe content despite the upstream filters, it's blocked from being shown or downloaded.
🛡️Human review for flagged content
Flagged generations are queued for human review. Confirmed violations result in credit forfeit, account warning, and — for repeated cases — account suspension.
Your responsibilities as a user
- You must have legal rights to every image you upload — product photos, designs, brand assets, and creator references.
- You must not attempt to bypass safety filters via obfuscated prompts, image steganography, or coordinated abuse.
- You agree not to use AgentForge outputs to mislead consumers, impersonate brands you don't own, or violate consumer protection law.
- If you operate AgentForge on behalf of clients, you are responsible for ensuring their use complies with this policy.
Consequences of misuse
First offence
Generation blocked. Credits used in the attempt are forfeit. Email warning sent to the account holder.
Repeated abuse
Account suspended pending review. In severe cases (CSAM, identity-fraud rings, hate campaigns), we cooperate with Indian law enforcement and report under applicable provisions of the IT Act 2000 and DPDP Act 2023.
Report misuse
If you encounter content generated on AgentForge that violates this policy — or believe your likeness has been used without consent — email info@aiagentforge.in with as much detail as you can share (URL, screenshot, generation ID if visible). We acknowledge reports within 72 hours.
For broader privacy and data-rights questions, see our Data Protection page.