Does Sora, Kling or Runway Allow NSFW? Policies (2026)

People who search “does Kling allow NSFW” or “can Sora make adult videos” usually want one word. For the big names, that word is no. We read the official usage policy, terms or community guidelines of 13 mainstream AI video and image tools on 2026-10-25. Twelve ban explicit adult content in plain words. One, Grok Imagine from xAI, allows a limited amount of 18+ content and bans sexual depictions of real people.

This guide gives each tool’s rule, a short quote from the company’s own page, and what the company says happens if you try. It ends with where adult AI video is allowed instead.

Which mainstream AI tools allow NSFW content?

Tool	Explicit adult content	What the policy says	What happens if you try	Last checked
Sora (OpenAI)	No	Guardrails “block unsafe content before it’s made—including sexual material”	Prompts and outputs checked across frames and audio. App closed April 26, 2026; API closed September 24, 2026	2026-10-25
Veo (Google)	No	Bans content “created for the purpose of pornography or sexual gratification”	Not stated on the policy page; exceptions possible for educational, documentary, scientific or artistic use	2026-10-25
Runway	No	Prohibits “Sexually explicit content, including sex toys and certain fetish content” and adult nudity	Account suspension for any violation; appeal by email	2026-10-25
Kling	No	Bans content “depicting sexual acts” in text, images or video	Restrictions on content generation; fees already paid not refunded	2026-10-25
Pika	No	Aims to be “PG-13 and family-friendly”; terms list pornographic content as prohibited	Suspension or termination; no refund after termination for breach	2026-10-25
Luma Dream Machine	No	“Do not request or generate content that is sexually explicit or intended for adults only”	Account flagged, limited or suspended; repeat NSFW triggers blocked	2026-10-25
Hailuo (MiniMax)	No	Bans content that would “Promote sexually explicit or pornographic material”	Content removed; access suspended or terminated	2026-10-25
Midjourney	No	“Do not create or attempt to create gore or adult content.”	Some prompts blocked automatically; warning, time-out or ban	2026-10-25
DALL-E / ChatGPT images (OpenAI)	No	Assistant “should not generate erotica, depictions of illegal or non-consensual sexual activities”	Request refused; narrow exceptions for scientific, historical, news or artistic context	2026-10-25
Adobe Firefly	No	Users may not “Create pornographic material or explicit nudity”	Prompts reviewed by automated and manual methods; accounts can be disabled	2026-10-25
Leonardo	No	Bans content with “explicit pornography”; NSFW blocked “by default”	Flagged prompt returns a content moderation error; suspension, then termination	2026-10-25
Ideogram	No	Bans output that is “sexually explicit” (section 6.2)	Content screened and removed; account suspended, disabled or deleted	2026-10-25
Grok Imagine (xAI)	Limited	Terms refer to “18+ adult content”; bans “Depicting likenesses of persons in a pornographic manner”	Action against the account, up to suspension or termination	2026-10-25

“Limited” means the company’s own pages allow some adult content but set clear limits. “Not stated” means we found no wording on that point on the official page.

How we compared

We read each company’s current usage policy, acceptable use policy, terms of service or community guidelines on 2026-10-25, and we recorded the update date the page showed. Where a company publishes two documents, such as Pika’s terms and its acceptable use policy, we read both and used the stricter rule.

The “explicit adult content” column answers one question: does the company’s own wording let an adult user make sexually explicit images or video? “No” means the page bans it. “Limited” means the page allows some 18+ content with limits. We did not send prompts to probe the filters, so the “what happens if you try” column reports what each company says it does, not what we saw.

We quote no more than 15 words from each source. Every page is listed in the sources at the end of this report. If a policy changes after our check, the company’s page wins.

Video tools: Sora, Veo, Runway, Kling, Pika, Luma and Hailuo

Sora (OpenAI)

OpenAI’s launch post for the Sora app, dated September 30, 2025, says its guardrails check both prompts and outputs across multiple video frames and audio transcripts, and that they block sexual material. The point is now mostly historical. OpenAI’s help center says the Sora web and app experiences were discontinued on April 26, 2026, and the Sora API on September 24, 2026. Users could export their content before OpenAI deletes the data.

Veo (Google)

Veo runs inside Google’s Gemini app, whose policy guidelines list video generation among Gemini features. Google’s Generative AI Prohibited Use Policy, last modified December 17, 2024, bans sexually explicit content made for pornography or sexual gratification. Gemini’s guidelines give “pornography or erotic content” as an example of what Gemini should not make. Google keeps room for exceptions on educational, documentary, scientific or artistic grounds, but no page we read describes an adult setting.

Runway

Runway’s Usage Policy, last updated March 6, 2026, has its own section on sexually explicit content and nudity. It prohibits sexually explicit content, adult nudity and attempts to create non-consensual intimate imagery. It also says Runway limits some suggestive content, including underwear or lingerie. Runway reserves the right to suspend an account for any violation, and users who think a suspension was wrong can appeal by email.

Kling

Kling’s Community Guidelines, released and effective on September 8, 2026, list generating or publishing content depicting sexual acts as prohibited conduct, in any form: text, images or video. They also ban promoting pornographic services. Penalties include restrictions on content generation, and the guidelines say fees already paid are not refunded.

Pika

Pika’s Terms of Service say it tries to keep the service PG-13 and family-friendly, and they list pornographic and obscene content as prohibited. Pika’s separate Acceptable Use Policy adds rules against non-consensual sexual content and deepfake pornography. Pika can suspend or terminate an account, gives no refund after termination for breach, and does not let a terminated user open a new account without written consent.

Luma Dream Machine

Luma’s Content Moderation Policy, last updated April 23, 2026, has a “No NSFW Content” rule for prompts and outputs that are sexually explicit or intended for adults only. It is the clearest enforcement statement in this group: attempts can get an account flagged, limited or suspended, and accounts that collect NSFW moderation triggers are blocked.

Hailuo (MiniMax)

The Hailuo terms, last modified August 19, 2026, cover hailuoai.video and the related Hailuo sites. Their content standards ban material that is obscene or indecent, and content that promotes sexually explicit or pornographic material. Hailuo can remove content and suspend or terminate access. The terms also set a minimum age of 16, which helps explain why the service keeps adult content out.

Image tools: Midjourney, ChatGPT, Firefly, Leonardo and Ideogram

Midjourney

Midjourney’s Community Guidelines tell users not to create gore or adult content. They define adult content broadly: nudity, sexual organs, sexualized imagery and fetishes. Midjourney blocks some text and image inputs automatically, and it notes that a prompt that is not blocked is not automatically allowed. Moderators can warn users, give a time-out or block them. Misusing the Editor to make NSFW or sexualized images can lead to suspension or a ban without refund.

DALL-E and ChatGPT images (OpenAI)

OpenAI’s Model Spec, dated October 29, 2026, governs ChatGPT’s text and visual output. It says the assistant should not generate erotica or depictions of illegal or non-consensual sexual activities, except in scientific, historical, news, artistic or similar contexts. OpenAI’s Usage Policies also ban non-consensual intimate content and any sexual content involving minors. The Model Spec still says OpenAI is exploring erotica “in age-appropriate contexts”. OpenAI announced an adult mode for verified adults in October 2025, but PCWorld, citing the Financial Times, reported on March 26, 2026 that it was put on hold indefinitely. We found no OpenAI page that says it has launched. For the text rules of ChatGPT, Grok, Gemini and Claude as chat assistants, see our guide to AI assistants and NSFW.

Adobe Firefly

Adobe’s Generative AI User Guidelines, last updated May 15, 2026, forbid using Firefly to create pornographic material or explicit nudity. Adobe says prompts and results may be reviewed by automated and manual methods, and that it can disable accounts for any violation.

Leonardo

Leonardo’s Terms of Service, last updated January 19, 2026, ban content that includes explicit pornography. Its API documentation says the API blocks NSFW image generation by default, like the web app, and that a flagged prompt returns an error such as “content moderation filter: nude”. The API also flags generated images that contain NSFW material. Leonardo can suspend access and, if the issue is not resolved within 14 days, terminate the account.

Ideogram

Ideogram’s Terms of Service, last revised August 14, 2024, ban user output that is sexually explicit in section 6.2. Ideogram may screen content with automated tools, remove it, and suspend, disable or delete accounts.

Grok Imagine (xAI): the one limited exception

xAI’s legal pages now appear under the name SpaceXSI. Its consumer Terms of Service, last updated October 12, 2026, say that some features and prompts can produce dialogue with sexual situations. They describe 18+ adult content that users can limit in Settings, and age assurance for users in Australia. That is why we mark Grok Imagine as “Limited” and not “No”.

The limits are firm. xAI’s Acceptable Use Policy, effective August 14, 2026, bans undressing or nudifying real persons, altering a real person’s likeness into an intimate or sexual context, and depicting anyone’s likeness in a pornographic manner. Violations can lead to action against the account, up to suspension or termination. On January 16, 2026, California’s Attorney General sent xAI a cease and desist letter demanding that it stop creating and distributing non-consensual sexual deepfakes and child sexual abuse material. We found no official xAI page that defines what Grok Imagine’s Spicy mode allows, so we do not describe it.

Why mainstream tools block adult content

Four reasons show up in the companies’ own pages and in public records.

1. App stores. Most of these tools ship mobile apps. Apple’s App Store Review Guidelines (1.1.4) reject apps with overtly sexual or pornographic material. A general-audience app that makes explicit content risks its listing.
2. Young users. Hailuo accepts users from age 16. Pika aims to be PG-13. OpenAI said the Sora app had stronger protections for teens, including limits on mature output. A service built for a mixed-age audience keeps adult content out by default.
3. Deepfakes and legal risk. Every policy here bans sexual depictions of real people without consent. The California Attorney General’s January 2026 letter to xAI shows the legal exposure when a tool produces sexual deepfakes.
4. Product focus. PCWorld, citing the Financial Times, reported that OpenAI shelved ChatGPT’s adult mode in March 2026 over user wellbeing concerns and the risk of minors getting access, as part of a wider shift toward coding and business users.

The result is the same at each company: filters on prompts and outputs, plus account penalties for repeat attempts. Trying to get around them can cost you the account, and in Kling’s and Pika’s cases, the money already paid.

Where to make adult AI video instead

If you are an adult who wants to make adult video legally, use a service whose rules allow it, not a mainstream tool you would have to trick.

Nureta is our top pick for adult AI video. It turns a written story into a multi-scene video with sound of up to about 3 minutes, and it keeps the same characters across scenes. Your first 5-second clip is free when you sign up with Google, with no card. After that, a 5-second clip costs $1.29, with no subscription. For developers, the Nureta API goes up to 1080p and supports first and last frame control. Nureta allows adult content and bans minors (also fictional ones), non-consent, real people without consent and public figures. Our Nureta review covers it in full.

Know its limits before you pay. The app renders at 480p; higher resolutions are on the API. Payment is USDC on Base only. Purchases are final, except failed jobs, which are refunded. Videos are deleted after 30 days, so download what you want to keep.

For the full field, see our best NSFW AI video generators ranking, which compares adult video tools on length, sound, character consistency, cost per clip and privacy. If you want to start without paying, our free NSFW AI video generators report lists which ones give free video and how much.

Which one should you choose?

• You want adult video: skip Sora, Veo, Runway, Kling, Pika, Luma and Hailuo. All seven ban it. Start with Nureta and its free first clip.
• You want adult images: Midjourney, ChatGPT, Firefly, Leonardo and Ideogram ban explicit images. See our best NSFW AI image generators for tools that allow them.
• You use Grok Imagine: some 18+ content is possible, but never with a real person’s face or likeness, and xAI publishes no full definition of its Spicy mode.
• You want safe-for-work video: every mainstream tool here except the closed Sora works, as long as you stay inside its policy. Read the “what happens if you try” column before you push the limits, because several of these services do not refund paid plans after a ban.

Whatever tool you use, two rules hold everywhere: no minors, and no sexual content of real people without their consent.