How to Create YouTube Videos with AI

The landscape of digital video production has undergone a tectonic paradigm shift. The era of low-effort, fully automated content that dominated the early 2020s has collapsed under the weight of algorithmic fatigue and strict platform regulations. In its place, 2026 has established a new standard for content creation: a highly sophisticated, human-directed methodology that leverages artificial intelligence as an augmentation tool rather than a replacement for human creativity. For aspiring creators seeking viable faceless YouTube channel ideas, established digital media teams looking to scale operations, and enterprise marketers demanding high-fidelity outputs, understanding the nuances of the 2026 AI video production ecosystem is no longer optional—it is a prerequisite for survival and monetization.

This comprehensive analysis details the optimal technical workflow, policy compliance strategies, and economic frameworks required to build a sustainable, monetizable YouTube presence utilizing the most advanced generative AI tools available today. The strategic imperative is clear: the integration of artificial intelligence into the creative pipeline must be executed with precision, balancing computational efficiency with authentic storytelling.

The State of AI Video in 2026: Beyond the Hype

The proliferation of multimodal AI models capable of generating high-fidelity text, audio, and video has fundamentally altered the economics of attention. As the global generative AI market for content creation rapidly approaches an $80 billion valuation, the volume of synthetically generated media has surged. However, this explosion of content has forced platforms to enforce stringent quality and transparency standards, ending the viability of zero-effort publishing.

The Shift from "Automation" to "Augmentation"

The most critical strategic pivot for creators in 2026 is the transition from full automation to the "Cyborg Approach." Early iterations of AI video production relied heavily on hands-off automation: scripts generated by large language models fed directly into robotic text-to-speech generators and paired with automated, disjointed stock footage. This fully automated approach frequently resulted in generic, mass-produced content derisively termed "AI slop," which algorithmic distribution systems now actively suppress.

Top creators and industry analysts have publicly warned against this methodology. Notably, digital luminaries such as MrBeast (Jimmy Donaldson) and Marques Brownlee (MKBHD) have expressed profound concerns regarding the hyper-realism of AI video and its impact on the creator economy. Donaldson has noted that when AI videos become indistinguishable from normal videos, it poses a direct threat to millions of creators who rely on traditional production. The industry consensus dictates that there are only two viable paths forward: massive-scale content farms that face constant platform termination risks, or authentic human storytelling heavily augmented by AI production tools.

The "Cyborg Approach" champions the human-in-the-loop methodology. It treats artificial intelligence as a highly capable production assistant rather than an autonomous creator. By adhering to the H-Y-B-R-I-D protocol—keeping a Human in charge, Yielding programmable steps, setting Boundaries, Reviewing with provenance, Instrumenting workflows, and Disclosing usage—content creators can ensure narrative longevity, brand safety, and deep audience connection. This methodology relies on human taste and editorial judgment to curate, refine, and synthesize the raw outputs generated by the machines, effectively merging the speed of algorithms with the emotional resonance of human experience.

YouTube’s 2026 AI Policy Update (The "Disclosure" Era)

To protect the integrity of the platform and maintain user trust, YouTube has instituted sweeping policy changes regarding artificial intelligence. Understanding these parameters is vital for maintaining a YouTube monetization AI content policy-compliant channel, as the platform has grown increasingly aggressive in terminating networks that violate these transparency mandates.

First, YouTube introduced mandatory disclosure labels for synthetically generated media to combat misinformation and deceptive practices. Creators must use the "altered content" setting in YouTube Studio during the upload flow to disclose content that is meaningfully altered or synthetically generated when it appears highly realistic. This explicit disclosure requirement applies when a creator makes a real person appear to say or do something they did not do, alters footage of a real event or place, or generates a realistic-looking scene that did not actually occur. Failure to disclose realistic synthetic content can result in content removal, suspension from the YouTube Partner Program (YPP), or the application of a permanent, unremovable warning label by YouTube's internal moderation teams. Notably, obviously unrealistic content—such as highly stylized animation or someone riding a unicorn—and the use of AI for productivity tasks like scriptwriting or brainstorming do not require this explicit disclosure.

Second, on July 15, 2025, YouTube rebranded its "repetitious content" monetization policy to the "inauthentic content" policy. This regulation specifically targets mass-produced or repetitive content that provides little educational or entertainment value, such as templated slideshows with synthetic narration that lack human curation. Rene Ritchie, YouTube's Creator Liaison, clarified that this is not a blanket ban on artificial intelligence. He explicitly stated that YouTube welcomes creators using AI tools to enhance storytelling, and channels that use AI in their content remain eligible to monetize, provided they add significant original commentary, modification, or educational value to the final product. The policy targets the absence of human effort, not the presence of a specific technology.

This distinction is reflected heavily in audience behavior. Recent data indicates that the average YouTube video retains only 23.7% of its viewers, with a massive 55% drop-off occurring within the first 60 seconds. However, hybrid videos that blend AI features with human-led instruction or human-curated narratives consistently outperform fully AI-generated, automated content, as viewers continue to seek authentic human takeaways and relatable insights.

Phase 1: Pre-Production & Scripting (The Brain)

The foundation of any successful video lies in its pre-production architecture. Because audience retention determines algorithmic reach, the conceptualization and scripting phases cannot be relegated to generic AI prompting. To create YouTube videos with AI effectively, the initial inputs must be meticulously engineered.

Validating Viral Ideas with Predictive AI

Before a single word of a script is drafted, the conceptual viability of a video idea must be rigorously tested. The highest-performing creators utilize predictive AI tools to analyze market gaps, keyword demand, and competitor performance, ensuring that production resources are allocated exclusively to ideas with proven audience appetite.

Enterprise analytics platforms like VidIQ and TubeBuddy have integrated advanced AI features specifically designed for trend prediction and topic validation in 2026. VidIQ, for example, offers an "AI Coach with Deep Thinking" capability, which provides personalized video idea generation based on channel-specific weighted metrics. By analyzing algorithmic trends, format repetition, and topic clusters appearing across related niches, creators can pressure-test concepts to confirm high search intent before committing to the visual generation phase.

These predictive algorithms evaluate the underlying architecture of viral videos, differentiating between weighted scores (personalized to a specific channel's historical performance and subscriber base) and unweighted scores (general market demand). This analytical rigor is a fundamental component of modern(/youtube-seo-guide), allowing creators to adapt proven, data-backed formats to their specific niches rather than relying on intuition.

Using LLMs for High-Retention Scripting (Gemini 1.5/GPT-4o)

When utilizing Large Language Models (LLMs) such as Gemini 1.5 Pro or GPT-4o for scriptwriting, it is a critical error to treat the model as a simple essay generator. The Cyborg Approach requires highly structured prompt engineering that dictates narrative pacing, visual cues, and psychological retention hooks. If a prompt is too broad, the resulting text will feel generic, predictable, and distinctly "AI-written," leading to immediate viewer abandonment.

A highly effective prompt framework forces the AI to adopt the persona of a "Professional YouTube Script Writer" and adhere to specific algorithmic retention mechanics. For instance, a robust script prompt should enforce the "2-1-3-4 Protocol" for educational or listicle videos. This structural methodology dictates that the script must begin with the second strongest point to act as an immediate, high-value hook, followed by the weakest point to bridge the narrative, and escalating to the strongest point at the climax to ensure viewers watch until the very end.

A direct comparison of prompting strategies illuminates this difference:

The Suboptimal Prompt:

"Write a 10-minute YouTube script about the history of artificial intelligence."

This prompt will generate a dry, chronological essay. It lacks pacing instructions, visual integration, and audience engagement triggers.

The Optimized Prompt:

"Act as a professional YouTube scriptwriter specializing in high-retention, algorithmic content. Write an 8-minute script about the evolution of artificial intelligence for an audience of tech enthusiasts. Employ the 2-1-3-4 narrative protocol to structure the arguments. Include a bold claim in the first 10 seconds to serve as a pattern interrupt hook. Insert a curiosity gap every 60 seconds to prevent viewer drop-off. Provide the output in a dual-column format: the left column must contain highly specific B-roll suggestions, text-on-screen cues, and sound effect annotations, while the right column contains the spoken dialogue tailored for a conversational, energetic delivery."

By mandating visual variety—such as suggesting a new camera angle, graphic, or stock footage clip every few seconds—the LLM outputs a comprehensive production blueprint rather than a flat manuscript. This methodology ensures the script is inherently optimized for the visual medium before the first frame is ever generated.

Phase 2: Visual Generation & Stock Assets (The Eye)

The visual component of AI video generation has experienced exponential improvements, transitioning from surreal, morphing abstractions to photorealistic, temporally consistent footage. The 2026 market is dominated by a few foundational video models capable of rendering cinematic physics and complex environments. Selecting the appropriate AI video generator for YouTube requires a nuanced understanding of cost, resolution, and specific narrative capabilities.

Generative Video Models: Sora, Veo, and Runway Gen-4.5

The enterprise and prosumer markets rely heavily on three primary text-to-video models. A technical and economic comparison is critical for establishing a sustainable production budget, particularly for teams executing daily uploads.

The implications of these specifications are profound for workflow design. For high-volume content automation, cost-per-second dictates profitability. Runway Gen-4.5 provides the most economical high-fidelity output for teams capable of executing their own sound design and post-production Foley. The platform's advanced camera controls—allowing for precise manipulation of pan, tilt, and zoom movements—make it highly favored by professional editors.

Conversely, Google Veo 3.1 fundamentally alters the post-production pipeline by generating native, synchronized audio directly from the text prompt. It can interpret scenic context and produce well-aligned sounds, such as sizzling in a kitchen or footsteps on gravel, simultaneously with the video. This capability potentially offsets its higher compute costs by entirely eliminating the downstream audio engineering phase for environmental sounds. OpenAI's Sora 2 remains the benchmark for complex physics and longer-form narrative cohesion, though its lack of reliable native audio necessitates a heavier post-production lift.

Regarding hardware requirements, the democratization of these tools means that heavy local processing power is no longer a strict barrier to entry. While institutional researchers and highly specialized enterprise teams may utilize local multi-GPU arrays (e.g., multiple RTX 3090s or 4090s with 96GB+ of VRAM) to handle distributed inference and custom diffusion models , standard YouTube production relying on proprietary systems like Sora, Veo, or Runway is entirely cloud-based. Users require only a standard web browser and a reliable internet connection, as the intense computational load is handled server-side by the respective platform's infrastructure.

AI B-Roll vs. Stock Footage: When to Use Which

A pervasive aesthetic danger in 2026 is the "Uncanny Valley"—the unsettling psychological effect produced when artificial intelligence generates hyper-realistic human faces that exhibit minor, unnatural flaws or dead-eyed stares. Content creators are heavily advised to avoid generating tight close-ups of realistic human faces unless using highly specialized, enterprise-grade talking avatars.

Instead, generative AI is optimally deployed for cinematic B-roll, abstract concepts, fantastical environments, historical reenactments, and stylized animation where absolute physical reality is not the baseline expectation. Traditional stock footage remains structurally superior for establishing absolute real-world trust. If a video discusses a specific geographic location, a breaking news event, or a precise medical procedure, integrating verified stock footage anchors the video in reality, mitigating viewer skepticism and preventing the entire production from feeling like an algorithmic hallucination. The most successful channels operate on a hybrid model, using AI to visualize the impossible and stock footage to ground the factual.

Consistent Character Generation for Storytelling

The most significant technical hurdle overcome in late 2025 and early 2026 is character consistency—the ability to maintain the exact appearance, clothing, and facial structure of a generated character across disparate scenes, lighting conditions, and camera angles.

OpenAI's Sora addresses this via the "Character ID" feature. Users extract features from an initial video or image to obtain a unique identifier. By utilizing an @id tag in subsequent prompts, the system summons the exact character into entirely new environments, drastically reducing "character drift" and raising visual similarity across scenes from 70% to over 99%.

Similarly, Runway Gen-4.5 utilizes advanced "References," allowing creators to lock in character traits, styles, and settings via reference images. By combining a "fixed seed" workflow with detailed prompt structuring that meticulously describes the character's facial proportions, clothing, and the specific camera lens required, creators can direct an AI character through long-form cinematic sequences without the visuals breaking down or hallucinating new attire. This capability allows independent creators to build recognizable intellectual property and recurring characters without hiring human actors.

Phase 3: Audio Engineering & Voice Synthesis (The Voice)

Audio fidelity is often the primary subconscious differentiator between amateur "AI slop" and professional, monetizable augmentation. Viewers will often tolerate minor visual artifacts, but robotic, emotionless audio severely and immediately damages audience retention.

Cloning Your Own Voice vs. Stock AI Avatars

Voice synthesis in 2026 has moved far beyond the monotonous text-to-speech engines of the past. Platforms like ElevenLabs have evolved their v3 models into comprehensive "Audio Operating Systems," prioritizing performance, contextual understanding, and emotional direction over mere pronunciation.

The critical innovation in ElevenLabs v3 is the integration of "Audio Tags." Creators can embed bracketed cues directly within their scripts to command the AI’s emotional delivery and situational awareness, effectively acting as a digital voice director. These tags fall into several categories:

• Emotional States: Formatting text with [excited], [nervous], or [sorrowful] drastically alters the pitch and cadence.

• Human Reactions: Inserting [sighs], [laughs], or [gasps] injects unscripted, natural human sounds into the audio track.

• Delivery Direction: Directing the model with [whispers], [shouting], or [stammers] alters the volumetric energy of the output.

This allows an AI voice clone to perform complex psychological beats. For instance, prompting the system with I think someone's in the house. Stay quiet. forces the AI to dynamically drop its volume, alter its resonance, and inject natural breath sounds. This completely eliminates the robotic cadence that YouTube's "inauthentic content" algorithms routinely flag. For channel longevity and brand building,(/review-of-elevenlabs) data suggests that cloning the creator's own voice (a Professional Voice Clone) yields significantly higher audience trust and monetization safety compared to utilizing over-saturated, recognizable stock voices.

Sound Design: AI Music and Foley Effects

Background music and Foley effects are heavily scrutinized by YouTube's automated Content ID system. AI music platforms like Suno and Udio provide an immediate solution for original score generation, but they operate within complex legal frameworks in 2026.

Suno's 2026 Terms of Service explicitly bifurcate usage rights based on subscription tiers. Users operating on the Free plan are strictly limited to non-commercial use, rendering those tracks ineligible for YouTube monetization. Conversely, Pro and Premier paid plans include full commercial rights, shielding creators from direct platform claims when uploading to monetizable digital environments.

However, possessing commercial usage rights from the platform does not equate to owning the copyright under federal law. Under 2026 U.S. Copyright Office guidelines, raw, unmodified AI-generated audio cannot be fully copyrighted by the end-user. To establish legally defensible intellectual property ownership, creators must introduce "Meaningful Human Authorship"—such as adding original human vocals, playing live instruments over the AI track, or heavily remixing the multitrack stems in a Digital Audio Workstation. Relying exclusively on raw AI music outputs presents a minor risk of "slop filtering" by digital service providers utilizing forensic watermarking detection.

For audio cleanup and mixing, Adobe's AI audio enhancement tools are integrated directly into editors like Premiere Pro. These tools utilize advanced machine learning to isolate primary vocals from severe background noise, eliminate room echo, and intelligently equalize frequencies, ensuring that even voiceovers recorded on substandard smartphone microphones achieve professional studio-quality resonance.

Phase 4: The Edit & Packaging (The Assembly)

The assembly phase dictates the final pacing, structural integrity, and psychological impact of the video. In 2026, the editing paradigm has fundamentally shifted away from traditional timeline-centric trimming toward highly efficient text-based video editing.

Text-Based Editing: Descript and Premiere Pro AI

Text-based editing tools, heavily pioneered by platforms like Descript and subsequently integrated natively into industry standards like Adobe Premiere Pro, transcribe the source media and present it to the editor as a word-processor document.

The workflow is profoundly efficient: when an editor highlights and deletes a word, sentence, or paragraph in the transcript text, the software automatically executes a precise ripple delete on the corresponding video and audio frames in the underlying timeline. This workflow incorporates sophisticated AI-driven "silence removal" and "filler word cuts," instantly eliminating "ums," "ahs," and awkward pauses across hours of footage with a single click. For creators utilizing the Cyborg approach, this completely automates the mechanical drudgery of rough-cutting, saving hundreds of hours and allowing human editors to focus their cognitive bandwidth exclusively on narrative pacing, visual overlays, and advanced graphical transitions.

Automating Thumbnails with Midjourney & Canva Magic Studio

The packaging of the video—specifically the thumbnail and the title—acts as the ultimate gatekeeper to the content, directly controlling the Click-Through Rate (CTR). Standard, unedited stock photos have lost their efficacy in a highly competitive feed.

In 2026, creators utilizing AI-optimized thumbnails generated via prompting in Midjourney and refined with typography in Canva Magic Studio report staggering CTR improvements of 20% to 30%. AI tools allow for the rapid generation of up to 5x more thumbnail variations in the same timeframe, enabling robust A/B testing frameworks. By synthesizing highly customized, emotionally evocative visuals that perfectly match the video's context and the target audience's psychological triggers, creators can bypass the physical limitations of traditional photography. These productivity gains are substantial, with some content marketing teams reducing design effort by nearly 90% while simultaneously increasing viewer click-through metrics. Exploring a Guide to Midjourney Prompts is essential for mastering this visual synthesis.

The "Hook" Strategy: Using AI to Optimize the First 30 Seconds

With granular analytics indicating that over 55% of viewer drop-off occurs within the first minute of a video, optimizing the "Hook" is paramount for algorithmic success. Videos that establish a clear, compelling value proposition within the first 15 seconds experience an 18% higher retention rate at the critical one-minute mark.

Creators use AI editing assistants to analyze their transcripts and ensure the hook adheres to proven retention strategies, such as Pattern Interrupts (starting with a visually or audibly surprising statement), Question Hooks, or Result Previews. By mathematically ensuring that the highest-energy audio delivery and the most dynamic visual generation are front-loaded into this initial 30-second window, creators maximize their Average View Duration (AVD). A high AVD acts as the primary signal to the YouTube recommendation algorithm that the content is highly engaging, triggering broader organic distribution.

Monetization & Copyright: Keeping Your Channel Safe

Navigating the intersection of generative artificial intelligence and intellectual property law requires rigorous compliance mechanisms to avoid catastrophic channel demonetization or severe legal liability.

Target Question: Can you monetize AI videos on YouTube?

Yes, but it requires the presence of meaningful human creativity and strict adherence to platform guidelines. YouTube does not ban the use of AI tools; rather, the platform demonetizes "inauthentic" or "repetitive" content that lacks human value, curation, or editorial oversight.

• Monetization Safe Practices:

  • Adding original human commentary, critical analysis, or personal opinions to AI-assisted scripts.

  • Utilizing dynamic, varied editing with motion graphics and logical, human-directed pacing.

  • Using your own cloned voice or highly expressive, directed AI voices (e.g., ElevenLabs Audio Tags) to avoid a robotic cadence.

  • Disclosing realistic synthetic media via YouTube’s "Altered Content" tool during the upload process.

• Monetization Risky Practices:

  • Uploading fully automated, templated image slideshows with minimal visual variation.

  • Relying on flat, un-directed text-to-speech narrators that read scripts without emotional nuance.

  • Publishing mass-produced, volume-based content with no discernible human curation or editing effort.

  • Creating deceptive deepfakes of real individuals (even for parody) without mandatory platform disclosures.

The "Reused Content" Trap: How to Avoid It

The most frequent reason AI-driven channels face sudden suspension or demonetization is falling into the "reused content" or "inauthentic content" trap. YouTube's automated review systems meticulously flag channels that display high upload frequencies with minimal structural variation between videos, interpreting them as spam.

A comparative analysis of channels highlights this dichotomy perfectly. Channels that bulk-generate hundreds of automated Reddit-thread reading videos using identical AI voices, scraping public text, and placing it over static backgrounds are routinely demonetized under the July 2025 YPP update. YouTube views this as low-effort mass production.

Conversely, heavily monetized "AI storytelling" channels—such as those operating in the horror narration, historical documentary, or true crime niches—succeed by utilizing a highly curated hybrid model. For example, case studies of faceless AI storytelling channels reaching 100,000 subscribers in under 30 days reveal a common thread: they succeed not by minimizing effort, but by utilizing AI to exponentially elevate production value. These channels employ advanced AI animation software to visualize bespoke, human-edited scripts, driven by a consistent, highly directed narrative voice and robust sound design. The defining factor for monetization is creative transformation; the AI must serve to execute a uniquely human vision, not to replace the human entirely.

Copyrighting Your AI Work: What the Law Says in 2026

The legal landscape surrounding AI training data, intellectual property, and copyright infringement is highly volatile and actively evolving. The central debate revolves around the doctrine of "Fair Use"—specifically, whether AI companies can legally scrape and ingest copyrighted material from the open internet to train their foundational models without compensating the original human authors. While initial, localized court rulings occasionally favored AI developers under the premise of "transformative use," the legislative branch has begun imposing stringent, sweeping regulations to protect human creativity.

In 2026, the Copyright Labeling and Ethical AI Reporting (CLEAR) Act was introduced at the federal level to address these exact concerns. This legislation mandates that developers of generative AI models must submit a highly detailed summary to the U.S. Copyright Office identifying every single copyrighted work included in their training datasets. Crucially, the CLEAR Act requires developers to file this comprehensive notice 30 days prior to the commercial release of the AI platform, and retroactive notices are required for existing systems. Furthermore, state-level regulations like California's AB 2013 (effective January 2026) force AI developers to publicly disclose detailed information about their training data, actively mitigating the "black box" problem and opening the door for copyright infringement lawsuits. Additionally, initiatives like the C2PA alliance and Google's SynthID embed invisible cryptographic watermarks into generated media, allowing platforms to trace the exact origin of a file down to the pixel level.

For YouTube creators, this legislative shift means the provenance of the tools they use is becoming entirely transparent. If a creator utilizes an open-source AI video generator found to be built on illegally misappropriated datasets, their generated assets could face future legal scrutiny or automated platform takedowns. Consequently, enterprise creators and risk-averse marketers are migrating rapidly toward AI platforms that operate on strictly "Licensed Models"—systems trained exclusively on legally acquired, compensated, or public domain data, offering creators contractual indemnification against copyright lawsuits.

Step-by-Step Workflow: Creating Your First Video Today

To practically synthesize the complex tools and policy strategies outlined in this report, creators can execute the "One-Hour Video" Challenge. This workflow represents a streamlined implementation of the Cyborg Approach, moving from concept to published video with maximum efficiency and compliance.

1. Ideation & Validation: Utilize the AI predictive features within VidIQ or TubeBuddy to identify a high-demand, low-competition topic within your specific niche. Validate the concept against outlier videos to ensure inherent algorithmic demand.

2. Script Generation: Prompt an advanced LLM like Gemini 1.5 Pro or GPT-4o using the "Professional YouTube Script Writer" persona framework. Explicitly enforce the 2-1-3-4 narrative protocol, demanding specific visual cues in a dual-column format and an aggressive, high-value 15-second opening hook.

3. Voice Synthesis: Export the refined dialogue into ElevenLabs v3. Insert specific Audio Tags (e.g., [excited], [pauses], [whispering]) directly into the text to meticulously direct the pacing and emotional resonance of the AI voice, or utilize your own trained Professional Voice Clone for maximum brand authenticity.

4. Visual Generation: Break the script’s visual cues into individual, highly descriptive prompts. Feed these into Google Veo 3.1 (for shots requiring native, synchronized audio and Foley) or Runway Gen-4.5 (for scenes demanding complex camera movements). Utilize reference images and fixed seeds to ensure strict character and stylistic consistency across all generated clips.

5. Music & Audio Sweetening: Generate a commercial-safe background score using a paid tier of Suno or Udio to ensure legal compliance. Process the final voiceover file through Adobe Podcast Enhance to equalize the frequencies and achieve studio-grade resonance.

6. Text-Based Editing: Import all generated visual and audio assets into Premiere Pro or Descript. Utilize the text-based editor interface to instantly execute silence removal, cutting dead air and seamlessly aligning the generated B-roll footage with the corresponding transcript dialogue. Ensure the first 30 seconds are visually dense, changing angles or assets every 3 to 5 seconds.

7. Packaging: Generate 3 to 5 highly evocative thumbnail variations using Midjourney prompts. Import these into Canva Magic Studio to refine the typography, ensuring maximum contrast and readability on mobile devices to drive the Click-Through Rate. For broader business strategies, ensure the description contains appropriate links.

8. Compliance & Upload: Upload the finalized, highly curated video to YouTube. Crucially, navigate to the "Altered Content" setting in the YouTube Studio upload flow and check "Yes" if the video depicts realistic synthetic scenes or human approximations, ensuring absolute compliance with the 2026 monetization guidelines.

By strictly adhering to this sophisticated workflow, creators can harness the unprecedented scale and efficiency of artificial intelligence while actively preserving the critical human elements of storytelling, authenticity, and legal compliance required to thrive in the modern digital economy.