Join waitlist

AI Video News: April 2026 Update on Models, Funding, Platforms, and the Slop Backlash

April 27, 2026

The AI video category just lived through its strangest six months. OpenAI launched the most-hyped video app of the year and then killed it. Runway raised $315M while its Hollywood deal quietly fell apart. YouTube wiped 4.7 billion views. The EU is about to make synthetic video labels mandatory. Here's what actually happened — and what creators, marketers, and platforms should be doing about it as of late April 2026.

TL;DR — six stories shaping AI video right now

  1. OpenAI is shutting down the Sora app on 26 April 2026 and the Sora API on 24 September 2026. Sora 2 launched 30 September 2025 with synchronized audio, multi-shot consistency, and a TikTok-style social feed. Six months later, advocacy groups, family estates, and SAG-AFTRA forced OpenAI to pull the plug after a wave of nonconsensual deepfakes and "AI slop" videos of figures like Michael Jackson and Martin Luther King Jr.
  2. Google shipped Veo 3 Lite on 31 March 2026 at $0.05/second for 720p and $0.08/second for 1080p, opening up audio-enabled video generation via the Gemini API and AI Studio. Veo 4 is widely expected later in 2026 but has not been officially confirmed.
  3. Runway raised $315M in February 2026 at a $5.3B valuation, led by General Atlantic with Nvidia, Fidelity, AllianceBernstein, and Mirae participating. Gen-4.5 currently leads the Artificial Analysis text-to-video benchmark at 1,247 Elo. Luma Labs sits at a $4B valuation after a $900M Series C in November 2025.
  4. TikTok now auto-labels AI content using C2PA Content Credentials and watermark detection and has applied AI labels to over 1.3 billion videos. YouTube terminated or wiped 16 of the top 100 AI slop channels — 35 million subscribers and an estimated $9.8M in annual ad revenue gone — and the platform's updated "inauthentic content" policy now disqualifies mass-produced AI uploads from the Partner Program.
  5. The EU AI Act's Article 50 transparency rules go live 2 August 2026. Providers of generative video must mark outputs in a machine-readable format. Deployers must visibly label deepfakes. Fines reach €15M or 3% of global turnover.
  6. Hollywood is fighting two fronts. SAG-AFTRA's contract with the studios expires in June 2026 and the union is bargaining for a "Tilly tax" — named after the AI actress Tilly Norwood — that would price synthetic performers at human-equivalent rates. Lionsgate's high-profile Runway training deal has reportedly stalled because the studio's catalog is too small to train a usable model.

The rest of this post is the full story with receipts.

The model wars: who shipped what since fall 2025

Twelve months ago, "AI video" still meant five-second silent clips with melting faces. That's no longer where the frontier is.

OpenAI Sora 2 (30 September 2025). Sora 2 was the moment AI video crossed into mainstream usability. Synchronized dialogue and sound effects baked into a single generation. Sharper physics — a missed basketball rebounds off the backboard the way it should. Multi-shot consistency, where characters and environments persist across cuts. OpenAI paired it with a new iOS social app called Sora featuring a customizable feed, remixing, and a "Cameos" feature that let users insert themselves or friends into generations. The app expanded to Android in markets including the U.S., Canada, Japan, Korea, Thailand, Vietnam, and Taiwan.

Google Veo 3 and Veo 3 Lite. Veo 3 launched in May 2025 with native audio and longer durations. Google then released Veo 3 Lite on 31 March 2026 as a budget tier through the Gemini API and Google AI Studio, priced at $0.05 per second for 720p and $0.08 per second for 1080p video. Veo 4 has been heavily speculated for sometime in 2026 — with promises of higher resolution, better character consistency, and stronger camera control — but Google has not confirmed a release date as of late April 2026.

Runway Gen-4.5. Runway leads the Artificial Analysis text-to-video benchmark at 1,247 Elo points as of early 2026, on the back of what the company calls "world consistency" — characters, environments, and objects holding coherent across cuts without re-prompting. Gen-4.5 added native audio, long-form multi-shot generation, and editing tooling.

Kling 3.0 (5 February 2026). Kuaishou's Kling shipped native 4K output, a storyboard tool with per-shot camera and pacing control, and a single-pipeline lip-synced audio system. Kling 3.0 is the model most credited with making prosumer-grade AI advertising plausible.

MiniMax Hailuo 2.3 and ByteDance Seedance 2.0. MiniMax's Hailuo 2.3 builds on the Hailuo 02 release that ranked #2 on Artificial Analysis benchmarks in mid-2025 — beating Veo 3 at the time. The 2.3 update sharpens dynamic expression, physical action realism, and character micro-expressions. ByteDance's Seedance 2.0 is now broadly available through aggregator platforms.

Pika 2.5. Pika has carved out a niche with Pikaframes, a feature that lets creators upload a start image and an end image and generate the 1–10-second visual transition between them. Pika is currently valued at $470M after an $80M Series B, with $135M raised total since its April 2023 founding.

Higgsfield. The San Francisco-based Higgsfield is the closest thing the category has to a Switzerland — a multi-model platform integrating Sora 2, Veo 3, Kling 3.0, MiniMax Hailuo 02, ByteDance Seedance 2.0, and Wan 2.6. It's backed by Accel, Menlo Ventures, GFT Ventures, and AI Capital Partners at a valuation north of $1.3B.

Meta Movie Gen. Meta has moved Movie Gen out of research and into the consumer product surface. The model — a 30B parameter Diffusion Transformer that generates 16-second 1080p HD video with synchronized audio and music — is now rolling out across Instagram, Facebook, and Messenger. Meta is also reportedly developing a successor image and video model targeted for a 2026 release.

xAI Grok Imagine. xAI launched Grok Imagine 1.0 on 2 February 2026 with 10-second 720p video and dramatically improved audio. The company's stated 2026 roadmap targets 30-minute video generation by year-end, with full-length films targeted for 2027.

The takeaway: there are now at least eight credible video model families, and the gap between the leader (Runway Gen-4.5) and the pack is small enough that most professional creators are running multi-model pipelines rather than committing to one provider.

Funding: the money is back in 2026

After a flat 2024, AI video funding doubled. Global funding for AI-related video companies in 2025 hit $3.08 billion, up 94.6% from $1.58 billion in 2024. The 2026 pace is accelerating.

  • Runway closed a $315M Series E in February 2026 at a $5.3B valuation, led by General Atlantic with participation from Nvidia, Fidelity, AllianceBernstein, and Mirae. The company also launched a $10M fund and a Builders program for early-stage AI startups in March 2026.
  • Luma AI raised a $900M Series C announced in November 2025 at a $4B valuation — one of the largest single rounds in the category to date.
  • Pika Labs sits at $470M after its $80M Series B (June 2024), with $135M raised cumulatively.
  • Higgsfield is over $1.3B in valuation, backed by Accel, Menlo Ventures, GFT Ventures, and AI Capital Partners.

Capital is concentrating. The video category looks increasingly like a small set of well-funded foundation-model labs (Runway, Luma, Pika, the big-tech entrants) and a longer tail of distribution-and-tooling layers built on top of them.

Platform policy: the rules just got real

Distribution is where AI video stops being a research curiosity and starts being a regulated medium.

TikTok

TikTok's 2026 AI content policy requires visible labels on all AI-generated visuals and audio depicting realistic people or scenes. The platform automatically applies labels using C2PA Content Credentials and invisible watermarking, and has now labeled over 1.3 billion videos as AI-generated. The detection system catches an estimated 35–45% of AI content automatically — up from just 18% in early 2024 — and TikTok closes the gap by requiring creator self-disclosure.

The enforcement teeth: after three unlabeled AI uploads, TikTok cuts account reach by roughly 60% for 30 days and pauses Creator Fund earnings.

YouTube

YouTube CEO Neal Mohan made "managing AI slop" a top platform priority for 2026. The numbers tell the story: as of early 2026, YouTube had terminated 11 channels and wiped content from 6 more among the top 100 AI slop accounts — 35 million subscribers, 4.7 billion lifetime views, and an estimated $9.8M in annual ad revenue erased. A March 2026 New York Times investigation found that roughly 40% of videos recommended to children on YouTube and YouTube Kids appeared to be AI slop.

The updated "inauthentic content" policy disqualifies mass-produced, templated, or machine-made content with no human creative input from the YouTube Partner Program. AI-assisted content where a creator's voice and editorial judgment are central remains monetizable. Channels using AI to produce near-identical uploads at scale now face manual review.

Instagram and Facebook

Meta unified AI content policies across Instagram and Facebook in February 2026. Both platforms now require in-post AI disclosure as a separate tag (similar to paid partnership labels), distinct from caption text.

The cross-platform reality: a single AI video may need a TikTok label, a Meta tag, a YouTube disclosure, and — soon — an EU-mandated watermark. One-size-fits-all compliance is over.

Hollywood: the Tilly tax, the Lionsgate stall, and Disney's Sora bet

The studio side of the AI video story has been, in a word, messy.

The Lionsgate-Runway training deal. In September 2024, Lionsgate became the first major studio to publicly partner with Runway, licensing its 20,000-title catalog to train a custom video model. By late 2025, the project was reportedly stalling. Sources told The Wrap and others that the Lionsgate catalog "is too small to create a model" — and that even Disney's catalog would be too small. Eighteen months in, the deal has produced storyboarding and pre-vis use cases but no flagship feature work.

Disney's $1B OpenAI investment. In December 2025, Disney announced a three-year licensing agreement to become "the first major content licensing partner on Sora," paired with a $1 billion investment into OpenAI. The Sora app shutdown announced in March 2026 doesn't kill that deal — the API stays live until 24 September 2026 and the underlying Sora 2 model continues — but it complicates the consumer-facing distribution story Disney bought into.

Netflix and Disney quietly using Runway. Reporting in early 2026 confirmed that Netflix and Disney have both been using Runway's tools internally — for storyboarding, VFX, and pre-visualization — even as the Lionsgate showcase project struggled. This is the pattern across studios: AI video is a craft tool inside the production pipeline, not yet a substitute for it.

SAG-AFTRA's 2026 contract and the Tilly tax. The union's contract with the AMPTP expires in June 2026, and AI is again the dominant bargaining issue. The headline ask: a so-called "Tilly tax," named after AI actress Tilly Norwood — the synthetic performer created by Eline Van der Velden whose talent-agency outreach in late September 2025 triggered industry-wide backlash. The Tilly tax would impose a fee on the use of synthetic performers in film and TV, designed to make AI characters cost roughly the same as human actors.

SAG-AFTRA has continued to expand its AI protection portfolio elsewhere. In its 2025 Commercials Contract, the union secured a requirement that advertisers and agencies obtain SAG-AFTRA permission before authorizing third parties to use commercial material to train AI systems. The union also filed an unfair labor practice charge against Llama Productions in 2025 over an AI-generated James Earl Jones voice for Darth Vader in Fortnite, and publicly condemned the use of AI replicas of Tom Cruise and Brad Pitt in viral Seedance demonstrations in early 2026.

Regulation: the EU AI Act lands on synthetic video in 99 days

If you generate, publish, or distribute AI video into the EU, the calendar matters.

Article 50 of the EU AI Act becomes enforceable on 2 August 2026. The obligations split across two roles:

  • Providers of generative AI systems (the labs and tooling companies) must mark synthetic audio, image, video, and text outputs in a machine-readable format. The technical solutions must be effective, interoperable, robust, and reliable — read: real watermarks and provenance metadata, not optional flags.
  • Deployers (anyone using AI professionally to generate or manipulate content) must clearly disclose deepfakes. The draft Code of Practice specifies modality-specific disclosure: persistent visual indicators and opening disclaimers for live video, visible labels for recorded video and images, audible disclaimers for audio.

The European Commission's draft Code of Practice on marking and labelling AI-generated content went out for consultation in early 2026, drew 187 written submissions, and is expected to be finalized by June 2026. It promotes a multilayered approach combining visible disclosures with invisible techniques such as C2PA-style content credentials and watermarking. The draft proposes a standardized EU-wide "AI" icon for synthetic content.

Penalties: up to €15 million or 3% of global annual turnover, whichever is higher, for Article 50 non-compliance. The omnibus simplification package that delayed parts of the AI Act in March 2026 did not touch the Article 50 deadline.

The U.S. picture is patchier — no federal synthetic media law, but state-level deepfake bans and political-ad disclosure rules continue to multiply. SAG-AFTRA has bipartisan congressional support for a federal NO FAKES-style anti-deepfake bill but no enacted federal statute.

What creators and marketers should do in the next 90 days

A practical checklist for the period between now and August 2026 enforcement:

For creators publishing AI video on social platforms: - Always self-disclose AI content at upload. TikTok's three-strike rule cuts reach by 60% for 30 days. YouTube's inauthentic-content policy can pull you out of the Partner Program. Meta requires in-post tags separate from captions. - If you've been running an AI-only channel, audit it against YouTube's mass-produced criteria. The 16 channels YouTube wiped were running templated, identical AI uploads with no creator presence. Add commentary, voice, and editorial perspective — or expect to lose monetization.

For marketers running AI video ads: - Assume any creative shipped into the EU after 2 August 2026 needs a visible AI label, machine-readable provenance metadata, and a documented compliance process. The fines start at €15M. - Track the final EU Code of Practice in June 2026 and align tooling — most major model providers are now embedding C2PA credentials by default.

For developers and product builders: - The model gap is small enough that locking into one provider is risky. Multi-model pipelines (Runway + Veo 3 Lite + Hailuo, for example) are the dominant pattern. - Watch the Sora API sunset on 24 September 2026 if you've built on top of it. OpenAI has not announced a successor video API as of late April 2026.

For studios and production companies: - The Lionsgate-Runway lesson is that catalog-scale training is harder than the press releases suggested. Production-pipeline AI (storyboarding, pre-vis, VFX) is delivering value today; AI feature generation is not. - SAG-AFTRA negotiations through June 2026 will reshape what synthetic-performer use costs. Plan budgets assuming a Tilly tax materializes in some form.

Bottom line

AI video in April 2026 is in the same awkward growth phase social media was in around 2010: the technology is real, the audience is real, the money is real, and the rulebook is being written in public, often badly. Sora 2's six-month rise and fall is the cautionary tale — a frontier-model launch that became a content-moderation crisis fast enough to take down the app. Runway's $315M raise and Luma's $4B valuation are the counterweight: serious capital is betting that the production-tool layer of AI video is durable.

The defensive posture is the same whether you're a creator, a marketer, or a studio: label your synthetic content, track the EU calendar, and assume the platforms will keep tightening monetization rules around mass-produced AI output. The offensive posture is to use the model gap — Runway, Veo, Kling, Hailuo, Sora 2 (while it lasts) — as a craft toolkit, not a content factory.

The headlines will keep moving. The compliance work and the craft work are what carry over.

Sources and further reading

On this page

Text Link
Text Link

Use our Free Forever Plan

Create and post one short video every day for free, and grow faster.

Try OpusClip
Try OpusClip

AI Video News: April 2026 Update on Models, Funding, Platforms, and the Slop Backlash

The AI video category just lived through its strangest six months. OpenAI launched the most-hyped video app of the year and then killed it. Runway raised $315M while its Hollywood deal quietly fell apart. YouTube wiped 4.7 billion views. The EU is about to make synthetic video labels mandatory. Here's what actually happened — and what creators, marketers, and platforms should be doing about it as of late April 2026.

TL;DR — six stories shaping AI video right now

  1. OpenAI is shutting down the Sora app on 26 April 2026 and the Sora API on 24 September 2026. Sora 2 launched 30 September 2025 with synchronized audio, multi-shot consistency, and a TikTok-style social feed. Six months later, advocacy groups, family estates, and SAG-AFTRA forced OpenAI to pull the plug after a wave of nonconsensual deepfakes and "AI slop" videos of figures like Michael Jackson and Martin Luther King Jr.
  2. Google shipped Veo 3 Lite on 31 March 2026 at $0.05/second for 720p and $0.08/second for 1080p, opening up audio-enabled video generation via the Gemini API and AI Studio. Veo 4 is widely expected later in 2026 but has not been officially confirmed.
  3. Runway raised $315M in February 2026 at a $5.3B valuation, led by General Atlantic with Nvidia, Fidelity, AllianceBernstein, and Mirae participating. Gen-4.5 currently leads the Artificial Analysis text-to-video benchmark at 1,247 Elo. Luma Labs sits at a $4B valuation after a $900M Series C in November 2025.
  4. TikTok now auto-labels AI content using C2PA Content Credentials and watermark detection and has applied AI labels to over 1.3 billion videos. YouTube terminated or wiped 16 of the top 100 AI slop channels — 35 million subscribers and an estimated $9.8M in annual ad revenue gone — and the platform's updated "inauthentic content" policy now disqualifies mass-produced AI uploads from the Partner Program.
  5. The EU AI Act's Article 50 transparency rules go live 2 August 2026. Providers of generative video must mark outputs in a machine-readable format. Deployers must visibly label deepfakes. Fines reach €15M or 3% of global turnover.
  6. Hollywood is fighting two fronts. SAG-AFTRA's contract with the studios expires in June 2026 and the union is bargaining for a "Tilly tax" — named after the AI actress Tilly Norwood — that would price synthetic performers at human-equivalent rates. Lionsgate's high-profile Runway training deal has reportedly stalled because the studio's catalog is too small to train a usable model.

The rest of this post is the full story with receipts.

The model wars: who shipped what since fall 2025

Twelve months ago, "AI video" still meant five-second silent clips with melting faces. That's no longer where the frontier is.

OpenAI Sora 2 (30 September 2025). Sora 2 was the moment AI video crossed into mainstream usability. Synchronized dialogue and sound effects baked into a single generation. Sharper physics — a missed basketball rebounds off the backboard the way it should. Multi-shot consistency, where characters and environments persist across cuts. OpenAI paired it with a new iOS social app called Sora featuring a customizable feed, remixing, and a "Cameos" feature that let users insert themselves or friends into generations. The app expanded to Android in markets including the U.S., Canada, Japan, Korea, Thailand, Vietnam, and Taiwan.

Google Veo 3 and Veo 3 Lite. Veo 3 launched in May 2025 with native audio and longer durations. Google then released Veo 3 Lite on 31 March 2026 as a budget tier through the Gemini API and Google AI Studio, priced at $0.05 per second for 720p and $0.08 per second for 1080p video. Veo 4 has been heavily speculated for sometime in 2026 — with promises of higher resolution, better character consistency, and stronger camera control — but Google has not confirmed a release date as of late April 2026.

Runway Gen-4.5. Runway leads the Artificial Analysis text-to-video benchmark at 1,247 Elo points as of early 2026, on the back of what the company calls "world consistency" — characters, environments, and objects holding coherent across cuts without re-prompting. Gen-4.5 added native audio, long-form multi-shot generation, and editing tooling.

Kling 3.0 (5 February 2026). Kuaishou's Kling shipped native 4K output, a storyboard tool with per-shot camera and pacing control, and a single-pipeline lip-synced audio system. Kling 3.0 is the model most credited with making prosumer-grade AI advertising plausible.

MiniMax Hailuo 2.3 and ByteDance Seedance 2.0. MiniMax's Hailuo 2.3 builds on the Hailuo 02 release that ranked #2 on Artificial Analysis benchmarks in mid-2025 — beating Veo 3 at the time. The 2.3 update sharpens dynamic expression, physical action realism, and character micro-expressions. ByteDance's Seedance 2.0 is now broadly available through aggregator platforms.

Pika 2.5. Pika has carved out a niche with Pikaframes, a feature that lets creators upload a start image and an end image and generate the 1–10-second visual transition between them. Pika is currently valued at $470M after an $80M Series B, with $135M raised total since its April 2023 founding.

Higgsfield. The San Francisco-based Higgsfield is the closest thing the category has to a Switzerland — a multi-model platform integrating Sora 2, Veo 3, Kling 3.0, MiniMax Hailuo 02, ByteDance Seedance 2.0, and Wan 2.6. It's backed by Accel, Menlo Ventures, GFT Ventures, and AI Capital Partners at a valuation north of $1.3B.

Meta Movie Gen. Meta has moved Movie Gen out of research and into the consumer product surface. The model — a 30B parameter Diffusion Transformer that generates 16-second 1080p HD video with synchronized audio and music — is now rolling out across Instagram, Facebook, and Messenger. Meta is also reportedly developing a successor image and video model targeted for a 2026 release.

xAI Grok Imagine. xAI launched Grok Imagine 1.0 on 2 February 2026 with 10-second 720p video and dramatically improved audio. The company's stated 2026 roadmap targets 30-minute video generation by year-end, with full-length films targeted for 2027.

The takeaway: there are now at least eight credible video model families, and the gap between the leader (Runway Gen-4.5) and the pack is small enough that most professional creators are running multi-model pipelines rather than committing to one provider.

Funding: the money is back in 2026

After a flat 2024, AI video funding doubled. Global funding for AI-related video companies in 2025 hit $3.08 billion, up 94.6% from $1.58 billion in 2024. The 2026 pace is accelerating.

  • Runway closed a $315M Series E in February 2026 at a $5.3B valuation, led by General Atlantic with participation from Nvidia, Fidelity, AllianceBernstein, and Mirae. The company also launched a $10M fund and a Builders program for early-stage AI startups in March 2026.
  • Luma AI raised a $900M Series C announced in November 2025 at a $4B valuation — one of the largest single rounds in the category to date.
  • Pika Labs sits at $470M after its $80M Series B (June 2024), with $135M raised cumulatively.
  • Higgsfield is over $1.3B in valuation, backed by Accel, Menlo Ventures, GFT Ventures, and AI Capital Partners.

Capital is concentrating. The video category looks increasingly like a small set of well-funded foundation-model labs (Runway, Luma, Pika, the big-tech entrants) and a longer tail of distribution-and-tooling layers built on top of them.

Platform policy: the rules just got real

Distribution is where AI video stops being a research curiosity and starts being a regulated medium.

TikTok

TikTok's 2026 AI content policy requires visible labels on all AI-generated visuals and audio depicting realistic people or scenes. The platform automatically applies labels using C2PA Content Credentials and invisible watermarking, and has now labeled over 1.3 billion videos as AI-generated. The detection system catches an estimated 35–45% of AI content automatically — up from just 18% in early 2024 — and TikTok closes the gap by requiring creator self-disclosure.

The enforcement teeth: after three unlabeled AI uploads, TikTok cuts account reach by roughly 60% for 30 days and pauses Creator Fund earnings.

YouTube

YouTube CEO Neal Mohan made "managing AI slop" a top platform priority for 2026. The numbers tell the story: as of early 2026, YouTube had terminated 11 channels and wiped content from 6 more among the top 100 AI slop accounts — 35 million subscribers, 4.7 billion lifetime views, and an estimated $9.8M in annual ad revenue erased. A March 2026 New York Times investigation found that roughly 40% of videos recommended to children on YouTube and YouTube Kids appeared to be AI slop.

The updated "inauthentic content" policy disqualifies mass-produced, templated, or machine-made content with no human creative input from the YouTube Partner Program. AI-assisted content where a creator's voice and editorial judgment are central remains monetizable. Channels using AI to produce near-identical uploads at scale now face manual review.

Instagram and Facebook

Meta unified AI content policies across Instagram and Facebook in February 2026. Both platforms now require in-post AI disclosure as a separate tag (similar to paid partnership labels), distinct from caption text.

The cross-platform reality: a single AI video may need a TikTok label, a Meta tag, a YouTube disclosure, and — soon — an EU-mandated watermark. One-size-fits-all compliance is over.

Hollywood: the Tilly tax, the Lionsgate stall, and Disney's Sora bet

The studio side of the AI video story has been, in a word, messy.

The Lionsgate-Runway training deal. In September 2024, Lionsgate became the first major studio to publicly partner with Runway, licensing its 20,000-title catalog to train a custom video model. By late 2025, the project was reportedly stalling. Sources told The Wrap and others that the Lionsgate catalog "is too small to create a model" — and that even Disney's catalog would be too small. Eighteen months in, the deal has produced storyboarding and pre-vis use cases but no flagship feature work.

Disney's $1B OpenAI investment. In December 2025, Disney announced a three-year licensing agreement to become "the first major content licensing partner on Sora," paired with a $1 billion investment into OpenAI. The Sora app shutdown announced in March 2026 doesn't kill that deal — the API stays live until 24 September 2026 and the underlying Sora 2 model continues — but it complicates the consumer-facing distribution story Disney bought into.

Netflix and Disney quietly using Runway. Reporting in early 2026 confirmed that Netflix and Disney have both been using Runway's tools internally — for storyboarding, VFX, and pre-visualization — even as the Lionsgate showcase project struggled. This is the pattern across studios: AI video is a craft tool inside the production pipeline, not yet a substitute for it.

SAG-AFTRA's 2026 contract and the Tilly tax. The union's contract with the AMPTP expires in June 2026, and AI is again the dominant bargaining issue. The headline ask: a so-called "Tilly tax," named after AI actress Tilly Norwood — the synthetic performer created by Eline Van der Velden whose talent-agency outreach in late September 2025 triggered industry-wide backlash. The Tilly tax would impose a fee on the use of synthetic performers in film and TV, designed to make AI characters cost roughly the same as human actors.

SAG-AFTRA has continued to expand its AI protection portfolio elsewhere. In its 2025 Commercials Contract, the union secured a requirement that advertisers and agencies obtain SAG-AFTRA permission before authorizing third parties to use commercial material to train AI systems. The union also filed an unfair labor practice charge against Llama Productions in 2025 over an AI-generated James Earl Jones voice for Darth Vader in Fortnite, and publicly condemned the use of AI replicas of Tom Cruise and Brad Pitt in viral Seedance demonstrations in early 2026.

Regulation: the EU AI Act lands on synthetic video in 99 days

If you generate, publish, or distribute AI video into the EU, the calendar matters.

Article 50 of the EU AI Act becomes enforceable on 2 August 2026. The obligations split across two roles:

  • Providers of generative AI systems (the labs and tooling companies) must mark synthetic audio, image, video, and text outputs in a machine-readable format. The technical solutions must be effective, interoperable, robust, and reliable — read: real watermarks and provenance metadata, not optional flags.
  • Deployers (anyone using AI professionally to generate or manipulate content) must clearly disclose deepfakes. The draft Code of Practice specifies modality-specific disclosure: persistent visual indicators and opening disclaimers for live video, visible labels for recorded video and images, audible disclaimers for audio.

The European Commission's draft Code of Practice on marking and labelling AI-generated content went out for consultation in early 2026, drew 187 written submissions, and is expected to be finalized by June 2026. It promotes a multilayered approach combining visible disclosures with invisible techniques such as C2PA-style content credentials and watermarking. The draft proposes a standardized EU-wide "AI" icon for synthetic content.

Penalties: up to €15 million or 3% of global annual turnover, whichever is higher, for Article 50 non-compliance. The omnibus simplification package that delayed parts of the AI Act in March 2026 did not touch the Article 50 deadline.

The U.S. picture is patchier — no federal synthetic media law, but state-level deepfake bans and political-ad disclosure rules continue to multiply. SAG-AFTRA has bipartisan congressional support for a federal NO FAKES-style anti-deepfake bill but no enacted federal statute.

What creators and marketers should do in the next 90 days

A practical checklist for the period between now and August 2026 enforcement:

For creators publishing AI video on social platforms: - Always self-disclose AI content at upload. TikTok's three-strike rule cuts reach by 60% for 30 days. YouTube's inauthentic-content policy can pull you out of the Partner Program. Meta requires in-post tags separate from captions. - If you've been running an AI-only channel, audit it against YouTube's mass-produced criteria. The 16 channels YouTube wiped were running templated, identical AI uploads with no creator presence. Add commentary, voice, and editorial perspective — or expect to lose monetization.

For marketers running AI video ads: - Assume any creative shipped into the EU after 2 August 2026 needs a visible AI label, machine-readable provenance metadata, and a documented compliance process. The fines start at €15M. - Track the final EU Code of Practice in June 2026 and align tooling — most major model providers are now embedding C2PA credentials by default.

For developers and product builders: - The model gap is small enough that locking into one provider is risky. Multi-model pipelines (Runway + Veo 3 Lite + Hailuo, for example) are the dominant pattern. - Watch the Sora API sunset on 24 September 2026 if you've built on top of it. OpenAI has not announced a successor video API as of late April 2026.

For studios and production companies: - The Lionsgate-Runway lesson is that catalog-scale training is harder than the press releases suggested. Production-pipeline AI (storyboarding, pre-vis, VFX) is delivering value today; AI feature generation is not. - SAG-AFTRA negotiations through June 2026 will reshape what synthetic-performer use costs. Plan budgets assuming a Tilly tax materializes in some form.

Bottom line

AI video in April 2026 is in the same awkward growth phase social media was in around 2010: the technology is real, the audience is real, the money is real, and the rulebook is being written in public, often badly. Sora 2's six-month rise and fall is the cautionary tale — a frontier-model launch that became a content-moderation crisis fast enough to take down the app. Runway's $315M raise and Luma's $4B valuation are the counterweight: serious capital is betting that the production-tool layer of AI video is durable.

The defensive posture is the same whether you're a creator, a marketer, or a studio: label your synthetic content, track the EU calendar, and assume the platforms will keep tightening monetization rules around mass-produced AI output. The offensive posture is to use the model gap — Runway, Veo, Kling, Hailuo, Sora 2 (while it lasts) — as a craft toolkit, not a content factory.

The headlines will keep moving. The compliance work and the craft work are what carry over.

Sources and further reading

Creator name

Creator type

Team size

Channels

linkYouTubefacebookXTikTok

Pain point

Time to see positive ROI

About the creator

Don't miss these

How Audacy Drove 1B+ Views by Taking a Tech-Forward Approach to Radio with OpusClip
No items found.

How Audacy Drove 1B+ Views by Taking a Tech-Forward Approach to Radio with OpusClip

How creators are earning 10M+ views in 1 month using video clipping
No items found.

How creators are earning 10M+ views in 1 month using video clipping

The Diary of a CEO: Scaling to 2M Subscribers with a Clips Strategy
No items found.

The Diary of a CEO: Scaling to 2M Subscribers with a Clips Strategy

AI Video News: April 2026 Update on Models, Funding, Platforms, and the Slop Backlash

No items found.
No items found.

Boost your social media growth with OpusClip

Create and post one short video every day for your social media and grow faster.
Try OpusClip

Related blogs

Success story

How OpusClip saves marketing agencies 40 hours monthly and boosts productivity 8X

Rebecca Xu
Head of Product Marketing at OpusClip
Success story

How OpusClip helps marketing agencies boost revenue by 148%

Rebecca Xu
Head of Product Marketing at OpusClip
Success story

Valuetainment Gained 512K New Subscribers in 90 Days Using OpusClip

Rebecca Xu
Head of Product Marketing at OpusClip
See more

AI Video News: April 2026 Update on Models, Funding, Platforms, and the Slop Backlash

April 27, 2026
15

The AI video category just lived through its strangest six months. OpenAI launched the most-hyped video app of the year and then killed it. Runway raised $315M while its Hollywood deal quietly fell apart. YouTube wiped 4.7 billion views. The EU is about to make synthetic video labels mandatory. Here's what actually happened — and what creators, marketers, and platforms should be doing about it as of late April 2026.

TL;DR — six stories shaping AI video right now

  1. OpenAI is shutting down the Sora app on 26 April 2026 and the Sora API on 24 September 2026. Sora 2 launched 30 September 2025 with synchronized audio, multi-shot consistency, and a TikTok-style social feed. Six months later, advocacy groups, family estates, and SAG-AFTRA forced OpenAI to pull the plug after a wave of nonconsensual deepfakes and "AI slop" videos of figures like Michael Jackson and Martin Luther King Jr.
  2. Google shipped Veo 3 Lite on 31 March 2026 at $0.05/second for 720p and $0.08/second for 1080p, opening up audio-enabled video generation via the Gemini API and AI Studio. Veo 4 is widely expected later in 2026 but has not been officially confirmed.
  3. Runway raised $315M in February 2026 at a $5.3B valuation, led by General Atlantic with Nvidia, Fidelity, AllianceBernstein, and Mirae participating. Gen-4.5 currently leads the Artificial Analysis text-to-video benchmark at 1,247 Elo. Luma Labs sits at a $4B valuation after a $900M Series C in November 2025.
  4. TikTok now auto-labels AI content using C2PA Content Credentials and watermark detection and has applied AI labels to over 1.3 billion videos. YouTube terminated or wiped 16 of the top 100 AI slop channels — 35 million subscribers and an estimated $9.8M in annual ad revenue gone — and the platform's updated "inauthentic content" policy now disqualifies mass-produced AI uploads from the Partner Program.
  5. The EU AI Act's Article 50 transparency rules go live 2 August 2026. Providers of generative video must mark outputs in a machine-readable format. Deployers must visibly label deepfakes. Fines reach €15M or 3% of global turnover.
  6. Hollywood is fighting two fronts. SAG-AFTRA's contract with the studios expires in June 2026 and the union is bargaining for a "Tilly tax" — named after the AI actress Tilly Norwood — that would price synthetic performers at human-equivalent rates. Lionsgate's high-profile Runway training deal has reportedly stalled because the studio's catalog is too small to train a usable model.

The rest of this post is the full story with receipts.

The model wars: who shipped what since fall 2025

Twelve months ago, "AI video" still meant five-second silent clips with melting faces. That's no longer where the frontier is.

OpenAI Sora 2 (30 September 2025). Sora 2 was the moment AI video crossed into mainstream usability. Synchronized dialogue and sound effects baked into a single generation. Sharper physics — a missed basketball rebounds off the backboard the way it should. Multi-shot consistency, where characters and environments persist across cuts. OpenAI paired it with a new iOS social app called Sora featuring a customizable feed, remixing, and a "Cameos" feature that let users insert themselves or friends into generations. The app expanded to Android in markets including the U.S., Canada, Japan, Korea, Thailand, Vietnam, and Taiwan.

Google Veo 3 and Veo 3 Lite. Veo 3 launched in May 2025 with native audio and longer durations. Google then released Veo 3 Lite on 31 March 2026 as a budget tier through the Gemini API and Google AI Studio, priced at $0.05 per second for 720p and $0.08 per second for 1080p video. Veo 4 has been heavily speculated for sometime in 2026 — with promises of higher resolution, better character consistency, and stronger camera control — but Google has not confirmed a release date as of late April 2026.

Runway Gen-4.5. Runway leads the Artificial Analysis text-to-video benchmark at 1,247 Elo points as of early 2026, on the back of what the company calls "world consistency" — characters, environments, and objects holding coherent across cuts without re-prompting. Gen-4.5 added native audio, long-form multi-shot generation, and editing tooling.

Kling 3.0 (5 February 2026). Kuaishou's Kling shipped native 4K output, a storyboard tool with per-shot camera and pacing control, and a single-pipeline lip-synced audio system. Kling 3.0 is the model most credited with making prosumer-grade AI advertising plausible.

MiniMax Hailuo 2.3 and ByteDance Seedance 2.0. MiniMax's Hailuo 2.3 builds on the Hailuo 02 release that ranked #2 on Artificial Analysis benchmarks in mid-2025 — beating Veo 3 at the time. The 2.3 update sharpens dynamic expression, physical action realism, and character micro-expressions. ByteDance's Seedance 2.0 is now broadly available through aggregator platforms.

Pika 2.5. Pika has carved out a niche with Pikaframes, a feature that lets creators upload a start image and an end image and generate the 1–10-second visual transition between them. Pika is currently valued at $470M after an $80M Series B, with $135M raised total since its April 2023 founding.

Higgsfield. The San Francisco-based Higgsfield is the closest thing the category has to a Switzerland — a multi-model platform integrating Sora 2, Veo 3, Kling 3.0, MiniMax Hailuo 02, ByteDance Seedance 2.0, and Wan 2.6. It's backed by Accel, Menlo Ventures, GFT Ventures, and AI Capital Partners at a valuation north of $1.3B.

Meta Movie Gen. Meta has moved Movie Gen out of research and into the consumer product surface. The model — a 30B parameter Diffusion Transformer that generates 16-second 1080p HD video with synchronized audio and music — is now rolling out across Instagram, Facebook, and Messenger. Meta is also reportedly developing a successor image and video model targeted for a 2026 release.

xAI Grok Imagine. xAI launched Grok Imagine 1.0 on 2 February 2026 with 10-second 720p video and dramatically improved audio. The company's stated 2026 roadmap targets 30-minute video generation by year-end, with full-length films targeted for 2027.

The takeaway: there are now at least eight credible video model families, and the gap between the leader (Runway Gen-4.5) and the pack is small enough that most professional creators are running multi-model pipelines rather than committing to one provider.

Funding: the money is back in 2026

After a flat 2024, AI video funding doubled. Global funding for AI-related video companies in 2025 hit $3.08 billion, up 94.6% from $1.58 billion in 2024. The 2026 pace is accelerating.

  • Runway closed a $315M Series E in February 2026 at a $5.3B valuation, led by General Atlantic with participation from Nvidia, Fidelity, AllianceBernstein, and Mirae. The company also launched a $10M fund and a Builders program for early-stage AI startups in March 2026.
  • Luma AI raised a $900M Series C announced in November 2025 at a $4B valuation — one of the largest single rounds in the category to date.
  • Pika Labs sits at $470M after its $80M Series B (June 2024), with $135M raised cumulatively.
  • Higgsfield is over $1.3B in valuation, backed by Accel, Menlo Ventures, GFT Ventures, and AI Capital Partners.

Capital is concentrating. The video category looks increasingly like a small set of well-funded foundation-model labs (Runway, Luma, Pika, the big-tech entrants) and a longer tail of distribution-and-tooling layers built on top of them.

Platform policy: the rules just got real

Distribution is where AI video stops being a research curiosity and starts being a regulated medium.

TikTok

TikTok's 2026 AI content policy requires visible labels on all AI-generated visuals and audio depicting realistic people or scenes. The platform automatically applies labels using C2PA Content Credentials and invisible watermarking, and has now labeled over 1.3 billion videos as AI-generated. The detection system catches an estimated 35–45% of AI content automatically — up from just 18% in early 2024 — and TikTok closes the gap by requiring creator self-disclosure.

The enforcement teeth: after three unlabeled AI uploads, TikTok cuts account reach by roughly 60% for 30 days and pauses Creator Fund earnings.

YouTube

YouTube CEO Neal Mohan made "managing AI slop" a top platform priority for 2026. The numbers tell the story: as of early 2026, YouTube had terminated 11 channels and wiped content from 6 more among the top 100 AI slop accounts — 35 million subscribers, 4.7 billion lifetime views, and an estimated $9.8M in annual ad revenue erased. A March 2026 New York Times investigation found that roughly 40% of videos recommended to children on YouTube and YouTube Kids appeared to be AI slop.

The updated "inauthentic content" policy disqualifies mass-produced, templated, or machine-made content with no human creative input from the YouTube Partner Program. AI-assisted content where a creator's voice and editorial judgment are central remains monetizable. Channels using AI to produce near-identical uploads at scale now face manual review.

Instagram and Facebook

Meta unified AI content policies across Instagram and Facebook in February 2026. Both platforms now require in-post AI disclosure as a separate tag (similar to paid partnership labels), distinct from caption text.

The cross-platform reality: a single AI video may need a TikTok label, a Meta tag, a YouTube disclosure, and — soon — an EU-mandated watermark. One-size-fits-all compliance is over.

Hollywood: the Tilly tax, the Lionsgate stall, and Disney's Sora bet

The studio side of the AI video story has been, in a word, messy.

The Lionsgate-Runway training deal. In September 2024, Lionsgate became the first major studio to publicly partner with Runway, licensing its 20,000-title catalog to train a custom video model. By late 2025, the project was reportedly stalling. Sources told The Wrap and others that the Lionsgate catalog "is too small to create a model" — and that even Disney's catalog would be too small. Eighteen months in, the deal has produced storyboarding and pre-vis use cases but no flagship feature work.

Disney's $1B OpenAI investment. In December 2025, Disney announced a three-year licensing agreement to become "the first major content licensing partner on Sora," paired with a $1 billion investment into OpenAI. The Sora app shutdown announced in March 2026 doesn't kill that deal — the API stays live until 24 September 2026 and the underlying Sora 2 model continues — but it complicates the consumer-facing distribution story Disney bought into.

Netflix and Disney quietly using Runway. Reporting in early 2026 confirmed that Netflix and Disney have both been using Runway's tools internally — for storyboarding, VFX, and pre-visualization — even as the Lionsgate showcase project struggled. This is the pattern across studios: AI video is a craft tool inside the production pipeline, not yet a substitute for it.

SAG-AFTRA's 2026 contract and the Tilly tax. The union's contract with the AMPTP expires in June 2026, and AI is again the dominant bargaining issue. The headline ask: a so-called "Tilly tax," named after AI actress Tilly Norwood — the synthetic performer created by Eline Van der Velden whose talent-agency outreach in late September 2025 triggered industry-wide backlash. The Tilly tax would impose a fee on the use of synthetic performers in film and TV, designed to make AI characters cost roughly the same as human actors.

SAG-AFTRA has continued to expand its AI protection portfolio elsewhere. In its 2025 Commercials Contract, the union secured a requirement that advertisers and agencies obtain SAG-AFTRA permission before authorizing third parties to use commercial material to train AI systems. The union also filed an unfair labor practice charge against Llama Productions in 2025 over an AI-generated James Earl Jones voice for Darth Vader in Fortnite, and publicly condemned the use of AI replicas of Tom Cruise and Brad Pitt in viral Seedance demonstrations in early 2026.

Regulation: the EU AI Act lands on synthetic video in 99 days

If you generate, publish, or distribute AI video into the EU, the calendar matters.

Article 50 of the EU AI Act becomes enforceable on 2 August 2026. The obligations split across two roles:

  • Providers of generative AI systems (the labs and tooling companies) must mark synthetic audio, image, video, and text outputs in a machine-readable format. The technical solutions must be effective, interoperable, robust, and reliable — read: real watermarks and provenance metadata, not optional flags.
  • Deployers (anyone using AI professionally to generate or manipulate content) must clearly disclose deepfakes. The draft Code of Practice specifies modality-specific disclosure: persistent visual indicators and opening disclaimers for live video, visible labels for recorded video and images, audible disclaimers for audio.

The European Commission's draft Code of Practice on marking and labelling AI-generated content went out for consultation in early 2026, drew 187 written submissions, and is expected to be finalized by June 2026. It promotes a multilayered approach combining visible disclosures with invisible techniques such as C2PA-style content credentials and watermarking. The draft proposes a standardized EU-wide "AI" icon for synthetic content.

Penalties: up to €15 million or 3% of global annual turnover, whichever is higher, for Article 50 non-compliance. The omnibus simplification package that delayed parts of the AI Act in March 2026 did not touch the Article 50 deadline.

The U.S. picture is patchier — no federal synthetic media law, but state-level deepfake bans and political-ad disclosure rules continue to multiply. SAG-AFTRA has bipartisan congressional support for a federal NO FAKES-style anti-deepfake bill but no enacted federal statute.

What creators and marketers should do in the next 90 days

A practical checklist for the period between now and August 2026 enforcement:

For creators publishing AI video on social platforms: - Always self-disclose AI content at upload. TikTok's three-strike rule cuts reach by 60% for 30 days. YouTube's inauthentic-content policy can pull you out of the Partner Program. Meta requires in-post tags separate from captions. - If you've been running an AI-only channel, audit it against YouTube's mass-produced criteria. The 16 channels YouTube wiped were running templated, identical AI uploads with no creator presence. Add commentary, voice, and editorial perspective — or expect to lose monetization.

For marketers running AI video ads: - Assume any creative shipped into the EU after 2 August 2026 needs a visible AI label, machine-readable provenance metadata, and a documented compliance process. The fines start at €15M. - Track the final EU Code of Practice in June 2026 and align tooling — most major model providers are now embedding C2PA credentials by default.

For developers and product builders: - The model gap is small enough that locking into one provider is risky. Multi-model pipelines (Runway + Veo 3 Lite + Hailuo, for example) are the dominant pattern. - Watch the Sora API sunset on 24 September 2026 if you've built on top of it. OpenAI has not announced a successor video API as of late April 2026.

For studios and production companies: - The Lionsgate-Runway lesson is that catalog-scale training is harder than the press releases suggested. Production-pipeline AI (storyboarding, pre-vis, VFX) is delivering value today; AI feature generation is not. - SAG-AFTRA negotiations through June 2026 will reshape what synthetic-performer use costs. Plan budgets assuming a Tilly tax materializes in some form.

Bottom line

AI video in April 2026 is in the same awkward growth phase social media was in around 2010: the technology is real, the audience is real, the money is real, and the rulebook is being written in public, often badly. Sora 2's six-month rise and fall is the cautionary tale — a frontier-model launch that became a content-moderation crisis fast enough to take down the app. Runway's $315M raise and Luma's $4B valuation are the counterweight: serious capital is betting that the production-tool layer of AI video is durable.

The defensive posture is the same whether you're a creator, a marketer, or a studio: label your synthetic content, track the EU calendar, and assume the platforms will keep tightening monetization rules around mass-produced AI output. The offensive posture is to use the model gap — Runway, Veo, Kling, Hailuo, Sora 2 (while it lasts) — as a craft toolkit, not a content factory.

The headlines will keep moving. The compliance work and the craft work are what carry over.

Sources and further reading

Ready to start streaming differently?

Opus is completely FREE for one year for all private beta users. You can get access to all our premium features during this period. We also offer free support for production, studio design, and content repurposing to help you grow.
Join the beta
Limited spots remaining

On this page

Text Link

Try OPUS today

A new way to create live streams. A magical way.
Try Opus Studio

About the Author

Derek Coleman

Derek Coleman is Senior Growth Marketer at OpusClip, where he leads SEO and Generative Engine Optimization strategy to help creators and brands grow their reach. He brings years of experience driving organic growth for SaaS and digital platforms.

Related blogs

Best Practice

How to Start Streaming in 2023: The Ultimate Guide

Rebecca Xu
Head of Product Marketing at OpusClip
Best Practice

7 Common Mistakes When Live Streaming and How to Avoid Them

Rebecca Xu
Head of Product Marketing at OpusClip
Opus Show

Opus Show with Ross Brand

Conor Eliot
Head of Creator Partnership at Opus
See more

Make your live stream your Magnum Opus

We and our vendors use technology that collects or records personal information about your use of our site so we can improve and personalize our products and services, for analytics and marketing, and to fulfill your requests. We may also share this information with marketing vendors, social media companies, and analytics partners, which may be considered "targeted advertising," "selling," or "sharing" under some state laws. By using our website, you acknowledge and agree to our Terms of Use and Privacy Policy.