HeyGen
AI avatar video platform for talking-head explainers, translation, and dubbing. 175+ languages. Growing newsroom adoption for localization — and growing deepfake concerns.
What should journalists know about HeyGen?
HeyGen generates synthetic talking-head videos from text scripts using AI avatars. Founded in 2020 by Joshua Xu and Wayne Liang (Carnegie Mellon alumni), the company hit $100M revenue in October 2025 after raising $60M at a $500M valuation from Benchmark in June 2024. The product is genuinely useful for localization: feed it a video in English, get back a lip-synced version in Korean, Arabic, or Spanish. For newsrooms producing explainers across multiple markets, this eliminates re-shoots. BBC, Reuters, and Al Jazeera have experimented with similar avatar tools for multilingual distribution. The deepfake problem is real. HeyGen's consent flow requires verbal confirmation with a spoken password before creating a personal avatar, and human moderators review flagged content. But the underlying technology is dual-use. The same system that produces a legitimate newsroom explainer can produce a convincing impersonation. HeyGen prohibits political content, violent content, and non-consensual avatars in its ToS, but enforcement is reactive. SOC 2 Type 2 certified. GDPR compliant. All data processed in the US on AWS. The company does not share user data with third parties. For journalism use: appropriate for clearly-labeled AI-generated explainers and translations. Not appropriate for anything that could be mistaken for footage of a real person saying real things.
Multilingual video explainers where you need the same segment in 10+ languages. Social media video clips for text-heavy newsrooms that lack video production capacity. Internal communications and training videos. Clearly-labeled AI presenter segments for newsletters and podcasts.
Anything that could be confused with real footage of real people. News segments where audience trust depends on authenticity of the presenter. Any use where AI generation is not disclosed. Political content. Investigative journalism where synthetic media undermines credibility.
Security & Privacy
Data is scrambled while being sent to their servers
Data is scrambled when stored on their servers
Where servers are located — affects which governments can request your data
Privacy policy summary
Account required. SOC 2 Type 2 certified. GDPR and CCPA compliant. Dedicated European DPO. Biometric data (face, voice) collected for avatar creation requires explicit consent. Data not shared with third parties beyond essential service providers (payment, cloud). Daily backups. Content moderation team reviews flagged outputs. Personal avatar creation requires verbal consent with spoken password verification.
How to protect yourself:
Always disclose AI-generated video to your audience — label it clearly. Do not create avatars of public figures or sources without explicit written consent. Use only for content types where synthetic presentation is editorially appropriate (explainers, translations, not news reporting). Review HeyGen's biometric privacy notice before uploading face/voice data. Keep copies of consent records if creating avatars of colleagues or talent. Monitor for unauthorized use of your likeness if you create a personal avatar.
HeyGen maintains SOC 2 Type 2 certification, GDPR compliance, and a structured consent flow for biometric data. The company's trust and safety team actively moderates content. The 'adequate' rating reflects the solid security infrastructure and privacy practices, balanced against the inherent dual-use risk of synthetic media technology and the absence of C2PA provenance on outputs. The consent mechanisms are better than most competitors, but the technology remains fundamentally capable of misuse.
Who Owns This
Known issues
Dual-use technology: the same system that produces legitimate explainers can produce convincing deepfakes. Consent enforcement is reactive — HeyGen cannot prevent all misuse before it happens. Credit-based pricing means unused credits expire monthly, creating unpredictable costs for irregular users. Voice cloning raises identity theft concerns if consent records are inadequate. No C2PA Content Credentials on output videos. Platform has been used in documented scam attempts involving cloned executive likenesses.
Pricing
Free: 1 credit (one 1-minute video). Creator: $29/month (200 credits, 1080p, voice cloning, 700+ avatars). Pro: $99/month (2,000 credits, 4K export, faster processing). Business: $149/month + $20/seat (shared credit pool, team collaboration, longer videos). Enterprise: custom pricing. Credits do not roll over. API pricing available separately.
This is an editorial assessment based on publicly available information as of 2026-04-11, using our published methodology. Independent security review is pending. Security posture can change at any time. This is not a guarantee of safety.
Something wrong or outdated? Report it.