A deepfake is audio, video, or imagery created or altered with artificial intelligence (AI) so that a person, event, or object appears or sounds different from reality. Deepfakes can be made for creative purposes, but they can also be used to impersonate people or spread misleading content. No single definition is universally accepted, so the term’s meaning depends partly on context.
What counts as a deepfake?
The U.S. Government Accountability Office (GAO) describes deepfakes as videos, audio, or images manipulated with AI, often to create, replace, or alter faces or synthesize speech. That broad description includes both media generated from scratch and changes to existing recordings. The National Security Agency (NSA) likewise defines the term broadly as multimedia synthetically created or manipulated using machine or deep-learning technology in its September 2023 guidance.
The UK Department for Science, Innovation and Technology’s 2025 report notes that there is no universally accepted definition. Its policy framing treats deepfakes as AI-generated or manipulated audio-visual material that misrepresents someone or something—including real or fictional people, events, or objects—and has potential to cause harm, regardless of intent. That is one useful framing, not a globally binding definition.
Deepfake versus ordinary editing
Not every edited photo or video is a deepfake. Cropping, color correction, or adding a conventional visual effect does not necessarily involve AI generation or manipulation. The label is most useful when AI is used to create or alter content in a way that changes what a viewer or listener might believe about who or what is represented.
#1 Best Overall
Examples across image, video, and audio
- Altered face or expression: AI changes a person’s face or expression in an existing video, or substitutes one face for another.
- Generated person or scene: AI produces an image or video depicting a person, event, or object that did not appear as shown in reality.
- Synthetic speech: AI generates a voice that sounds like a real person saying words they did not say.
These examples illustrate why deepfake is not just another word for fake video: the content may be a still image or audio recording, and manipulation can be subtle or extensive.
Why deepfakes are made—and when they can cause harm
Deepfakes can support creative effects in entertainment and commerce. They can also be used to impersonate people, create non-consensual sexual imagery, or attempt to influence elections. The GAO’s 2024 overview documents such harmful uses and warns about the limits of detection. The NSA and federal partners also describe risks to brands, finances, and public understanding when synthetic media spreads false claims about political, social, military, or economic issues.
Rank #2
Those examples establish that deepfakes can be misused; they do not show that every deepfake is malicious or how much of all deepfake content serves each purpose. The UK’s 2025 report on deepfake detection technology groups detection-service use cases into fraud prevention and cybersecurity; misinformation and narrative-manipulation detection; identity and age verification; reputation, brand protection, and social monitoring; content moderation; secure real-time communications; and national security and law enforcement. These are areas where services may be used, not evidence that every service performs equally well in them.
How to assess whether media might be a deepfake
Some detection systems look for facial or vocal inconsistencies, traces associated with AI generation, or color abnormalities. But a suspicious-looking detail is not proof, and a polished clip is not proof of authenticity. GAO says current detection methods have limited effectiveness in real-world conditions, can be evaded, and may not prevent a fake from spreading after it is identified.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Use a detector, if available, as one input—not as a final verdict. For consequential claims, check the source and context independently:
- Find the earliest available version. Check who first posted the media and whether a credible source provides the original recording or context.
- Look for independent confirmation. Search for reliable reporting or statements from people and organizations directly involved, rather than relying on reposts of the same clip.
- Compare what is claimed with what the media shows. A genuine recording can be presented with a false date, location, or explanation; authenticity of the file alone does not establish the surrounding claim.
- Treat automated results cautiously. A detector’s output can help guide further checking, but the available evidence does not support treating consumer detection tools as infallible.
Detection and authentication address different questions. Detection looks for signs that media may have been manipulated. Authentication approaches, such as watermarks, may provide information about provenance or alteration. Neither approach alone guarantees that a clip is true, complete, or being presented in context. Organizations face a broader cybersecurity and communications challenge: identifying suspicious media, preparing a response, and reducing the chance that an impersonation or false claim causes harm.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How widespread are deepfakes?
Reliable statistics on the frequency and impact of AI-generated fake content remain limited, according to the International AI Safety Report 2025. It reports that 43% of people aged 16 and older in the UK said they had seen at least one deepfake online in the previous six months; the figure was 50% among children aged 8–15. These are UK survey findings for those populations and that recall period—not global rates or a measure of how many deepfakes exist. They also do not establish that respondents could verify that the media they saw was a deepfake.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
