As artificial intelligence makes manipulated video, audio, and images easier to create, who should be responsible for deciding what is authentic, misleading, or dangerous?
Should the federal government regulate synthetic media? Should technology companies identify and remove it? Or should individuals be expected to evaluate the credibility of what they see and hear for themselves?
Before exploring this topic, I understood the general idea of deepfakes but was less familiar with the artificial intelligence used to produce them or the growing difficulty of distinguishing manipulated media from authentic content. The more I read, the more apparent it became that the danger does not depend entirely on whether a fake is technically sophisticated.
The Department of Homeland Security explains that the threat of deepfakes and synthetic media comes partly from our natural inclination to believe what we see. Manipulated content does not have to be perfect to influence opinions, spread confusion, or reinforce something an audience already wants to believe.
That is what makes the issue especially concerning.
Deepfakes and shallowfakes
Deepfakes generally use artificial intelligence to create or alter audiovisual content. A person’s face, voice, expressions, or movements can be reproduced in a way that makes it appear that they said or did something that never occurred.
Shallowfakes can be much simpler. They may involve selectively edited clips, changed playback speed, misleading captions, removed context, or unrelated audio placed over an existing video.
In some ways, shallowfakes may be just as effective because they require less technical expertise and can be produced quickly. The content only needs to appear believable long enough to be shared, repeated, and accepted by part of the audience.
Without the knowledge or tools needed to distinguish authentic media from manipulated content, we are approaching a serious intersection in how we determine what is real.
The role of technology companies
Large technology companies have several advantages in addressing manipulated media. Platforms such as Google, Facebook, Microsoft, and other technology providers have the technical infrastructure needed to analyze content, identify suspicious patterns, label altered media, and limit its distribution.
They can develop detection tools, strengthen content-authentication systems, preserve information about where media originated, and provide warnings when material may have been manipulated.
However, placing too much authority in the hands of large technology companies creates its own concerns.
Do we want a private company deciding what is true, false, misleading, satirical, or politically harmful? Who determines the standards? How transparent are the detection systems? What happens when the system makes a mistake?
A company may remove harmful material, but it may also suppress legitimate journalism, satire, creative work, or political criticism. Automated systems are not free from error, bias, or the priorities of the organizations that created them.
Technology companies should therefore play an important role, but their decisions should be transparent, reviewable, and open to appeal.
The role of government
The federal government also has a legitimate interest in addressing malicious synthetic media, especially when it involves fraud, election interference, threats, impersonation, national security, or reputational harm.
Government can establish legal definitions, penalties, disclosure requirements, and protections for people who are intentionally misrepresented. It can also fund research into detection, digital authentication, and public education.
Yet government involvement creates similar concerns.
Lawmakers and public officials may have political interests of their own. A regulation intended to limit deception could be used too broadly, applied unevenly, or interpreted in a way that restricts lawful speech.
Public trust is already divided. Depending on which political party or institution delivers the message, some people will accept the government’s conclusion while others will immediately reject it.
This means that government cannot simply announce that a piece of media is false and expect the matter to be settled.
Shared responsibility
The best approach is likely one of shared responsibility.
Government should establish clear legal protections against harmful impersonation, fraud, and malicious manipulation. Technology companies should develop detection and authentication tools, label questionable content, and provide transparent enforcement processes. News organizations and educators should help people understand how synthetic media works. Individuals must also develop stronger habits of verification.
No single group should have complete control over determining truth.
The issue is too important to be left entirely to private corporations, government agencies, or individual users acting alone.
The continuing role of human judgment
My thoughts on this issue are consistent with how I approach influential media more broadly: think, reason, and make decisions according to evidence and principle rather than simply adopting the opinions of others.
That does not mean ignoring expert analysis or rejecting institutional guidance. It means resisting the impulse to believe or share something immediately because it supports an existing opinion.
Before accepting a suspicious video or recording, we should ask:
- Where did it originate?
- Is the complete version available?
- Has it been reported by multiple reliable sources?
- Does the audio match the visible movements?
- Are there unexplained cuts, timing problems, or contextual gaps?
- Is the content designed primarily to provoke anger or fear?
- Is there independent evidence that the event occurred?
Even these questions will not always provide certainty, but they create a pause between exposure and belief.
That pause matters.
Podcast: Christian Nunes on Deepfakes (with Max Tegmark)
Regulation will not eliminate the technology
Regardless of how aggressively governments or companies regulate deepfakes, the technology will continue to exist. Malicious actors will find new methods, move to different platforms, and improve their ability to avoid detection.
This does not mean regulation is pointless. Laws, detection systems, and authentication tools can reduce harm and make deception more difficult. However, they cannot replace public awareness and critical thinking.
The long-term challenge is not merely detecting individual fakes. It is maintaining a society in which people still believe that reliable evidence exists.
If every image or recording can be dismissed as fake, authentic evidence can also lose its power. A person confronted with a real recording may claim that it was artificially generated. This creates what may be an even greater danger: not that everyone believes the false, but that no one believes anything.
Final thought
Deepfakes force us to reconsider how trust is created in a digital environment.
The federal government has a role in addressing criminal misuse and establishing protections. Technology companies have a role in detection, labeling, authentication, and limiting distribution. Journalists, educators, and researchers have a role in helping the public understand the technology.
Individuals also have a responsibility to slow down, verify claims, and resist sharing emotionally provocative content before determining whether it is credible.
The question is not simply who should police deepfakes.
The more difficult question is how government, technology companies, institutions, and individuals can share responsibility without giving any one group unchecked authority over what society is permitted to see and believe.