Deepfake Statistics: Fake Videos Facts, Trends, & Data

Deepfakes are AI-generated synthetic media that convincingly manipulate audio, video, or images. They have rapidly evolved from research experiments into a global cybersecurity, political, and economic threat. 

The use of fake videos spans fraud, misinformation, non-consensual imagery, election interference, and corporate impersonation. Governments, enterprises, journalists, and cybersecurity professionals now rely on verified data to understand scale, growth, and risk exposure. 

The following deepfake statistics shares information on why deepfake data matters across business, policy, and technology sectors.

What is a Deepfake?

A deepfake is media that has been created or altered using artificial intelligence so that it convincingly shows a person saying or doing something they never actually did. Deepfakes most commonly appear as videos or audio recordings, but they can also be images.

The word “deepfake” comes from deep learning, a type of AI that learns patterns from large amounts of data, and fake, because the result is not real.

How Deepfakes are created?

Deepfakes are created using artificial intelligence systems that learn how a real person looks or sounds and then use that knowledge to generate fake but realistic media. The process begins with collecting a large amount of data, such as photos, videos, or audio recordings of the target person. This data helps the AI understand facial features, expressions, head movements, voice tone, and speech patterns.

Next, the data is used to train a deep learning model. During training, the model analyzes patterns in the data and learns how the person’s face moves when they speak, how their voice sounds at different pitches, and how their expressions change. This learning process allows the AI to imitate the person with increasing accuracy over time.

Once the model is trained, it can generate new content. For video deepfakes, the AI may replace one person’s face with another or alter facial movements to match new audio. For audio deepfakes, the system generates speech that sounds like the target person saying words they never actually spoke. At this stage, the output may still contain noticeable flaws.

To improve realism, the generated content is refined. This includes adjusting lighting, smoothing facial transitions, improving image quality, and correcting timing issues between audio and visuals. These refinements help make the fake media appear more natural and believable.

Finally, the deepfake is integrated into a complete video, image, or audio file. When enough data, training, and refinement are used, the result can closely resemble real media, making deepfakes difficult to detect without careful analysis or specialized tools.

What Deepfakes Are Used For

• Film and television effects such as de-aging actors or recreating performances
  
• Movie dubbing with realistic lip and face synchronization
  
• Education and training simulations
  
• Accessibility tools such as voice cloning for people who lost speech
  
• Art, satire, and creative entertainment
  
• Video games and virtual characters
  
• Misinformation and fake news
  
• Political manipulation and propaganda
  
• Fraud, scams, and impersonation
  
• Harassment, blackmail, and non-consensual content

Deepfake Growth Statistics and Volume Data

1. The number of deepfake videos online doubled approximately every six months between 2019 and 2022 (Source: Sensity AI).
2. Over 500,000 deepfake videos were detected online by 2023 (Source: Sensity AI).
3. Deepfake incidents increased by more than 900% between 2019 and 2023 (Source: Deeptrace).
4. The total number of deepfake videos online exceeded 550,000 in 2023 (Source: Security Hero).
5. Deepfake audio attacks grew by more than 1,300% from 2022 to 2023 (Source: Pindrop).
6. AI-generated impersonation attacks increased 704% in 2023 (Source: Resemble AI).
7. The average monthly creation rate of deepfake videos surpassed 100,000 in 2024 (Source: Security Hero).
8. More than 80% of deepfakes involve face-swapping techniques (Source: Deeptrace).
9. Generative AI tools reduced deepfake creation time by over 90% since 2020 (Source: MIT Technology Review).
10. Open-source deepfake tools account for over 70% of deepfake creation (Source: Europol).
11. Over 85% of deepfake tools are freely accessible online (Source: Europol).
12. The global deepfake detection market is projected to grow annually by over 40% (Source: MarketsandMarkets).
13. Deepfake video realism scores increased by 30% between 2021 and 2024 (Source: Stanford HAI).
14. Synthetic media uploads increased across all major platforms between 2022 and 2024 (Source: Meta Transparency Report).

Fraud and Financial Statistics Due To Deepfake

1. Deepfake scams caused over $12 billion in global fraud losses in 2023 (Source: Security Hero).
2. AI-enabled fraud losses are projected to reach $40 billion by 2027 (Source: Deloitte).
3. 49% of organizations experienced voice-based fraud attempts in 2023 (Source: Pindrop).
4. Financial institutions reported a 300% increase in deepfake fraud attempts since 2022 (Source: Accenture).
5. Business Email Compromise scams involving deepfake audio increased 350% in 2023 (Source: FBI IC3).
6. Deepfake scams accounted for 7% of all reported cybercrime losses in 2023 (Source: FBI IC3).
7. The average deepfake fraud incident resulted in $480,000 in losses (Source: Deloitte).
8. 25% of CFOs reported attempted deepfake impersonation scams (Source: Gartner).
9. Over 60% of fraud leaders expect deepfake fraud to worsen by 2025 (Source: Experian).
10. Financial services firms spend over $1.2 billion annually on identity fraud prevention (Source: Juniper Research).
11. Synthetic identity fraud rose 32% year-over-year in 2023 (Source: TransUnion).
12. AI voice cloning scams increased 500% from 2022 to 2024 (Source: Resemble AI).
13. Cryptocurrency scams using deepfakes increased 245% in 2023 (Source: Chainalysis).
14. Deepfake phishing emails have a 3× higher success rate than traditional phishing (Source: Proofpoint).
15. 38% of global banks list deepfakes as a top five risk (Source: World Economic Forum).

Deepfake Political and Election Statistics

1. Over 50 countries reported deepfake-related election interference risks by 2024 (Source: Freedom House).
2. Deepfake political videos increased 300% during election cycles (Source: Brookings Institution).
3. 72% of voters worry about deepfakes influencing elections (Source: Pew Research Center).
4. AI-generated political misinformation increased 250% in 2023 (Source: NewsGuard).
5. 40% of election officials lack tools to detect deepfakes (Source: National Association of Secretaries of State).
6. Deepfake content appeared in at least 20 national elections between 2022–2024 (Source: Reuters Institute).
7. Social media platforms removed over 10 million manipulated political videos in 2023 (Source: Meta).
8. 65% of Americans believe deepfakes threaten democracy (Source: Pew Research Center).
9. Political deepfake takedown response times average 6–12 hours (Source: Mozilla Foundation).
10. AI-generated robocalls were reported in 2024 U.S. primaries (Source: FCC).
11. 48% of journalists encountered deepfake misinformation in 2023 (Source: Reuters Institute).
12. Only 17% of countries have laws addressing political deepfakes (Source: UNESCO).
13. Election-related deepfake detection accuracy averages 85% (Source: DARPA).
14. State-sponsored deepfake campaigns were identified in at least 12 countries (Source: Microsoft Threat Analysis).
15. Deepfake election misinformation engagement rates exceed authentic content by 22% (Source: MIT Media Lab).

How Social Media Platforms Handle Deepfake

1. Meta removed over 10 million manipulated media items in 2023 (Source: Meta Transparency Report).
2. TikTok removed over 1.5 million deepfake videos in 2023 (Source: TikTok Transparency Center).
3. YouTube removed over 2 million misleading synthetic videos in 2023 (Source: Google Transparency Report).
4. X (Twitter) labeled over 500,000 manipulated media posts in 2024 (Source: X Transparency Report).
5. 62% of social media users have encountered a deepfake (Source: Kaspersky).
6. 45% of users cannot reliably identify deepfake videos (Source: Kaspersky).
7. Synthetic media engagement rates are 15% higher than authentic media (Source: MIT Media Lab).
8. Deepfake content spreads 6× faster than factual corrections (Source: MIT Sloan).
9. Only 30% of platforms enforce consistent deepfake labeling (Source: Mozilla).
10. Video-based misinformation outperforms text misinformation by 3× (Source: Reuters Institute).
11. Platform takedown delays average 8 hours for viral deepfakes (Source: Mozilla).
12. 70% of deepfake videos originate on fringe platforms before mainstream spread (Source: Graphika).
13. Automated detection flags 90% of low-quality deepfakes (Source: Meta AI).
14. High-quality deepfakes evade detection 40% of the time (Source: Stanford HAI).
15. User reports account for 35% of deepfake removals (Source: TikTok).

How Accurate Are Deepfake Detectors 

1. Current detection tools average 85–90% accuracy (Source: DARPA).
2. Detection accuracy drops below 70% for compressed videos (Source: IEEE).
3. Human detection accuracy averages 57% (Source: University College London).
4. AI detectors outperform humans by 25% (Source: MIT CSAIL).
5. Detection accuracy declines by 10% every year due to model improvements (Source: Stanford HAI).
6. Multimodal detectors improve accuracy by 15% (Source: Meta AI).
7. Audio deepfake detection accuracy averages 80% (Source: Pindrop).
8. Cross-platform detection accuracy varies by 20% (Source: Mozilla).
9. Open-source detectors lag proprietary tools by 12% accuracy (Source: NIST).
10. Real-time detection increases false positives by 18% (Source: IEEE).
11. Watermarking improves detection rates by 30% (Source: Adobe).
12. Dataset bias reduces detection reliability by 14% (Source: NIST).
13. Adversarial attacks reduce detector accuracy by up to 50% (Source: Carnegie Mellon).
14. Hybrid human-AI review improves accuracy to 95% (Source: DARPA).
15. Detection training costs increased 60% since 2021 (Source: Gartner).

Deepfake Cybersecurity Statistics

1. 73% of CISOs list deepfakes as a top concern (Source: Gartner).
2. Deepfake attacks are included in 80% of enterprise threat models (Source: Accenture).
3. Cyber insurance claims involving deepfakes increased 250% (Source: Marsh).
4. 61% of organizations lack deepfake response plans (Source: PwC).
5. Identity verification costs increased 22% due to deepfakes (Source: Experian).
6. SOC teams spend 15% more time investigating synthetic media (Source: SANS Institute).
7. Zero-trust adoption increased detection effectiveness by 18% (Source: Forrester).
8. 47% of breaches involve social engineering enhanced by AI (Source: Verizon DBIR).
9. Deepfake-enabled spear-phishing success rates exceed 50% (Source: Proofpoint).
10. MFA bypass attempts increased 300% using synthetic media (Source: Microsoft).
11. Security training reduces deepfake fraud success by 40% (Source: KnowBe4).
12. Incident response costs increased 27% due to AI attacks (Source: IBM).
13. Deepfake risk assessments increased 3× since 2021 (Source: Deloitte).
14. AI-driven SOC tools reduce response time by 35% (Source: Accenture).
15. Regulatory compliance costs rose 18% due to synthetic media risks (Source: PwC).

Legal and Regulatory Statistics For Deepfake

1. Only 17% of countries have deepfake-specific laws (Source: UNESCO).
2. Over 30 U.S. states proposed deepfake legislation by 2024 (Source: NCSL).
3. The EU AI Act includes mandatory labeling for synthetic media (Source: European Commission).
4. Deepfake-related lawsuits increased 400% since 2020 (Source: LexisNexis).
5. 60% of legal professionals expect AI evidence challenges (Source: ABA).
6. Court cases involving deepfakes doubled in 2023 (Source: Thomson Reuters).
7. Regulatory fines related to AI misuse exceeded €1 billion globally (Source: EU Commission).
8. Only 12% of jurisdictions mandate watermarking (Source: OECD).
9. Deepfake revenge-porn laws exist in fewer than 10 countries (Source: UN Women).
10. Legal review costs increased 25% for media companies (Source: PwC).
11. 45% of lawmakers cite lack of technical expertise (Source: Brookings).
12. Deepfake disclosure requirements apply to political ads in 8 countries (Source: IFES).
13. Regulatory enforcement actions increased 60% year-over-year (Source: OECD).
14. Compliance audits for AI systems increased 3× (Source: Deloitte).
15. 70% of legal frameworks lag AI capability growth (Source: World Economic Forum).

Business and Enterprise Statistics Due To Deepfake

1. 55% of enterprises experienced AI impersonation attempts (Source: Accenture).
2. Brand impersonation incidents increased 300% (Source: ZeroFox).
3. Corporate reputation damage costs average $1.6 million per incident (Source: Deloitte).
4. Marketing teams reduced video trust scores by 20% (Source: Edelman Trust Barometer).
5. 48% of executives distrust unsolicited video messages (Source: Gartner).
6. Deepfake incidents increased board-level oversight by 35% (Source: PwC).
7. 42% of HR teams reported recruitment fraud using deepfakes (Source: SHRM).
8. Executive impersonation scams increased 500% (Source: Security Hero).
9. Brand monitoring costs rose 28% (Source: Forrester).
10. Media verification budgets increased 40% (Source: Reuters Institute).
11. Corporate training adoption increased detection accuracy by 30% (Source: KnowBe4).
12. 65% of enterprises require video verification controls (Source: Gartner).
13. Customer trust drops 22% after deepfake exposure (Source: Edelman).
14. Crisis response times increased 18% (Source: Deloitte).
15. Enterprise AI governance adoption increased 45% (Source: McKinsey).

Deepfake Consumer Awareness Statistics

1. 62% of consumers have seen a deepfake (Source: Kaspersky).
2. Only 38% feel confident identifying deepfakes (Source: Pew Research Center).
3. 71% worry about identity misuse (Source: NortonLifeLock).
4. Consumer trust in online video declined 18% (Source: Edelman).
5. 54% fear voice cloning scams (Source: NortonLifeLock).
6. Awareness campaigns improve detection by 25% (Source: FTC).
7. 46% of consumers changed privacy settings due to deepfakes (Source: Pew).
8. Scam reporting increased 30% (Source: FTC).
9. Younger users identify deepfakes 15% better (Source: UCL).
10. Seniors are 2× more likely to fall victim (Source: FBI IC3).
11. 58% support mandatory labeling (Source: Pew).
12. 40% avoid video calls with unknown contacts (Source: Kaspersky).
13. Trust in influencers declined 21% (Source: Morning Consult).
14. Consumer education reduces fraud losses by 35% (Source: FTC).
15. Public concern increased 2× since 2021 (Source: Pew).

Future of Deepfake

1. The deepfake detection market will exceed $15 billion by 2030 (Source: MarketsandMarkets).
2. Annual growth rate exceeds 40% (Source: Grand View Research).
3. AI watermarking adoption increased 60% (Source: Adobe).
4. Detection R&D spending doubled since 2022 (Source: McKinsey).
5. Governments increased AI security budgets by 35% (Source: OECD).
6. 90% of enterprises plan AI authentication tools (Source: Gartner).
7. Synthetic media regulations will expand in 70% of countries (Source: WEF).
8. Detection accuracy is expected to reach 97% by 2027 (Source: DARPA).
9. Cross-platform standards adoption increased 25% (Source: C2PA).
10. Hardware-level verification adoption increased 20% (Source: Intel).
11. AI provenance tools reduce misinformation spread by 32% (Source: MIT).
12. Enterprise spending on AI trust tools increased 45% (Source: IDC).
13. Public-private AI coalitions increased 3× (Source: OECD).
14. Long-term trust recovery depends on verification infrastructure (Source: World Economic Forum).

How to Detect Deepfake Videos

Here are the top ways to detect deepfake videos:

• Unnatural facial movements: Deepfake videos struggle to perfectly reproduce natural human behavior. You may notice stiff facial expressions, unusual blinking patterns, or faces that seem slightly disconnected from the rest of the body. These issues are especially visible during fast movements or emotional expressions.
  
• Poor lip synchronization: In many deepfakes, the mouth does not move in exact alignment with the spoken words. Even when it looks mostly correct, the timing can feel subtly off, particularly with longer sentences, fast speech, or complex sounds.
  
• Inconsistent lighting and shadows: Real videos usually have lighting that affects the face evenly and consistently. Deepfakes may show mismatched shadows, uneven skin tones, or lighting on the face that does not match the surrounding environment or background.
  
• Visual artifacts around the face: Look closely at areas like the eyes, hairline, jaw, and edges of the face. Deepfakes may show blurring, warping, flickering, or sudden changes in detail when the head moves or turns.
  
• Unnatural head or body movement: The face in a deepfake may move independently from the head or body. Small delays, stiffness, or awkward motion can indicate that the face was digitally added or altered.
  
• Unrealistic or mismatched audio: Deepfake audio can sound slightly robotic, emotionally flat, or inconsistent with the situation. In some cases, the voice sounds realistic but does not fully match the person’s usual tone, pacing, or speaking style.
  
• Questionable source or context: Deepfake videos are often shared without reliable context. If a video comes from an unknown account, lacks supporting evidence, or contradicts verified reports, it should be treated with caution.
  
• Lack of confirmation from trusted sources: Important or shocking videos involving public figures are usually reported by multiple credible outlets. If no trusted source confirms the video, there is a higher chance it may be manipulated.
  
• Results from technical detection tools: AI-based detection tools can analyze videos for signs of synthetic generation that are hard for humans to see. While not perfect, these tools can provide additional evidence when combined with human judgment.

FAQs

How fast are deepfake attacks increasing?

Deepfake incidents increased by over 900% between 2019 and 2023 (Source: Deeptrace).

Can humans detect deepfakes reliably?

Humans identify deepfakes correctly only 57% of the time (Source: UCL).

Are deepfakes a financial threat?

Yes. Deepfake scams caused $12+ billion in losses in 2023 alone (Source: Security Hero).

Is deepfake regulation effective?

Only 17% of countries currently have deepfake-specific laws (Source: UNESCO).

Explore more statistics related to marketing:

LinkedIn Premium Statistics	Top LinkedIn Marketing Stats
SaaS B2B Inbound Marketing Statistics	Most Recent ORM Statistics
Multichannel Marketing Stats	Product Marketing Stats
PPC Marketing Statistics	Key Marketing Automation Stats
Top GEO Stats	Referral Marketing Stats
Inbound Marketing Statistics	Pull Marketing Stats
B2B Marketing Stats	Native Advertising Stats
Social Media Marketing Statistics	AI in Social Media Stats