When anything can be faked who do we trust

[NovaBurst57]

TL;DR

• AI can now generate convincing images, voices and videos, so seeing and hearing are no longer reliable proof on their own
• Legal scholars Robert Chesney and Danielle Citron warned in 2019 of a liar's dividend, where real evidence can be dismissed as fake once fakes are common
• Psychology helps explain our vulnerability: research on the truth default and the illusory truth effect shows that people tend to believe what they hear and what they hear repeatedly
• Philosopher Onora O'Neill argued that what matters is not more trust but well placed trust, directed at people and institutions that deserve it
• Practical responses include verification habits, provenance technologies, prebunking and stronger relationships with reliable sources

The end of seeing is believing

For most of human history, seeing something with your own eyes or hearing a familiar voice was strong evidence. Photographs and recordings extended that trust across distance and time. Courts, newspapers and families relied on them. That foundation is now shaking

Generative AI can create realistic images of events that never happened, clone voices from short samples and produce videos of people saying things they never said. In 2024, an employee of engineering firm Arup in Hong Kong transferred about US$25 million after a video call in which colleagues, including a chief financial officer, were deepfakes. Smaller voice cloning scams target families every day. The technology is improving and becoming cheaper

This thread is not mainly about detection techniques, which another thread here covers. It asks a deeper question. When anything can be faked, who and what should we trust? How do our minds decide what to believe, and how can individuals and societies adapt? I hope members will share both personal experiences and philosophical views

How our minds handle truth

Psychologists have long studied how people decide what to believe. Communication researcher Timothy Levine's truth default theory, developed over many years and set out in a 2014 paper, proposes that people generally assume others are telling the truth unless something prompts suspicion. This default makes social life possible, because constant suspicion would be exhausting. It also makes us vulnerable to skilled deception

Repetition affects belief too. In 1977, psychologists Lynn Hasher, David Goldstein and Thomas Toppino reported what became known as the illusory truth effect: statements people hear repeatedly are rated as more likely to be true, regardless of whether they are. Later research has confirmed the effect many times. A deepfake shared widely may gain credibility simply by being seen often

Emotion plays a role as well. Content that provokes anger, fear or delight spreads faster and is scrutinised less. Deceptive media often exploits these reactions deliberately. Recognising our own emotional responses is one of the first steps towards more careful judgement

Group identity matters too. People tend to accept information that fits the views of groups they belong to and reject information that challenges them. Fakes that confirm existing beliefs are therefore especially persuasive. Recognising this tendency in ourselves is uncomfortable but useful

The liar's dividend

In a 2019 paper in the California Law Review, legal scholars Robert Chesney and Danielle Citron described a troubling side effect of deepfakes. As people learn that convincing fakes exist, those caught on genuine recordings can claim the evidence is fake. They called this the liar's dividend. The more common deepfakes become, the easier it is to dismiss real evidence

This effect may be more damaging than any individual fake. If people cannot agree on basic facts, public debate, journalism and justice become harder. Accountability depends on evidence that can be trusted. A world where everything can be denied is a world where powerful people face fewer consequences

It also affects ordinary people. A real photograph of harassment, a recording of a threat or video of an accident could be dismissed as AI generated. Victims may struggle to be believed. Protecting the credibility of genuine evidence is as important as exposing fakes

Trust, trustworthiness and dependence

Philosopher Onora O'Neill, in her 2002 BBC Reith Lectures titled A Question of Trust, argued that the goal is not simply more trust. Trust given blindly can be exploited. What matters is well placed trust, directed at people and institutions that are in fact trustworthy, and withheld from those that are not. Her ideas are especially relevant in an age of synthetic media

Philosopher John Hardwig made a related point in 1985 in a paper called Epistemic Dependence. He observed that most of what we know comes from trusting others, such as experts, teachers and reporters, because no individual can verify everything personally. Rational belief often depends on trusting the right sources. Deepfakes do not change that dependence, but they raise the stakes of choosing sources wisely

Together, these ideas suggest a shift in focus. Instead of asking whether a particular image is real, we can ask whether the source that shared it has earned our trust. Reputation, track record and transparency become more important when content itself is unreliable. Relationships with trustworthy institutions, journalists and communities matter more than ever

Where trust might move

If content alone can no longer prove much, trust may shift towards provenance, meaning where content comes from and how it has been handled. Technologies such as C2PA Content Credentials attach tamper evident records showing how media was created and edited. Some cameras, editing tools and platforms support them. If widely adopted, they could help distinguish authentic material from fabricated content

Trust may also shift towards relationships. In families, a pre agreed code word can confirm identity during a suspicious call. In businesses, callback procedures and dual approvals prevent fraud. These methods rely on shared knowledge and established channels rather than on how convincing a voice or face appears

Local knowledge regains value. Neighbours, community groups and local journalists often know what happened in a place better than distant viral posts. Supporting local information sources strengthens resilience. Communities with strong local networks may be harder to fool

Institutions could become more important as verifiers. Reputable news organisations, fact checkers and public bodies that verify content openly and explain their methods may gain trust. Those that cut corners may lose it quickly. The value of careful verification rises as fakes become cheaper

Building resilience

Researchers have explored ways to help people resist manipulation. Psychologists Sander van der Linden and Jon Roozenbeek, among others, have studied prebunking, which exposes people to weakened examples of manipulation techniques so they recognise them later. Their online game Bad News, released in 2018, let players take the role of a fake news creator. Studies found that such approaches can improve people's ability to spot manipulation, although effects vary and can fade over time

Media literacy education teaches habits such as checking sources, seeking corroboration and noticing emotional manipulation. Methods like Mike Caulfield's SIFT approach, which stands for stop, investigate the source, find better coverage and trace claims to their origin, are simple enough for anyone to use. Schools, libraries and community groups increasingly teach these skills. They work regardless of how good fakes become

Personal humility helps too. Everyone can be fooled, including experts. Being willing to say I am not sure, and to correct ourselves publicly when wrong, builds a healthier information culture. Communities that reward careful checking over speed are harder to manipulate

Should we become more sceptical

There is a balance to strike. Too little scepticism leaves us open to deception. Too much can lead to cynicism, where people believe nothing and dismiss everything they dislike as fake. That cynicism serves manipulators almost as well as gullibility does

Philosophers sometimes describe the ideal as calibrated trust: believing things in proportion to the evidence and the reliability of sources. Calibration is hard, but it is a skill that can be practised. It means holding some beliefs firmly, others tentatively and remaining open to revision. In an age of synthetic media, calibration may be one of the most important civic virtues

Connections to post quantum security

There is a technical link to the themes of this forum. Digital signatures underpin many provenance systems, software updates and identity checks. They prove that content or messages come from who they claim. Today most digital signatures use algorithms that a future quantum computer could break

If signatures could be forged, provenance systems and digital identity would lose their value, deepening the trust crisis. That is one reason the migration to post quantum signatures matters for society, not just for security teams. Trustworthy infrastructure supports trustworthy information. Q-Day preparation and the fight against misinformation are more connected than they first appear

Questions to discuss

• Have you ever been fooled by a fake image, video or voice, and what did you learn
• Who do you trust most for information today, and why
• Is the liar's dividend a bigger danger than deepfakes themselves
• Should platforms be required to label AI generated content
• How do we teach children to be sceptical without becoming cynical

Closing thoughts

When anything can be faked, trust cannot rest on appearances alone. Psychology explains why we are vulnerable, while philosophy reminds us that trust has always depended on choosing sources wisely. Provenance technologies, verification habits, prebunking and strong relationships offer ways forward. The challenge is to place our trust well, neither too freely nor not at all

Please share your experiences and views in the replies. Practical tips for families and communities are especially welcome. Thoughtful discussion is itself a small act of building trust


Further reading

• Robert Chesney and Danielle Citron, Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security, California Law Review, 2019
• Timothy Levine, Truth-Default Theory, Journal of Language and Social Psychology, 2014
• Lynn Hasher, David Goldstein and Thomas Toppino, Frequency and the conference of referential validity, Journal of Verbal Learning and Verbal Behavior, 1977
• Onora O'Neill, A Question of Trust, BBC Reith Lectures, 2002
• John Hardwig, Epistemic Dependence, Journal of Philosophy, 1985
• Jon Roozenbeek and Sander van der Linden, Fake news game confers psychological resistance against online misinformation, Palgrave Communications, 2019

[Leo50]

Most of all, the liar's dividend worries me. I have already heard people dismiss real news footage as AI simply because they disagreed with it. It happened twice in my family group chat last month. Nobody checked anything before dismissing it

Once that habit spreads, facts stop mattering. Arguments become pure tribalism. That is frightening for democracy

Provenance technology might help. But only if people care about it

[Highland Fatima]

Onora O'Neill's distinction between trust and trustworthiness is so useful. We should not be asking how to make people trust more. We should ask how to make institutions deserve trust. That reframing changes the whole debate

If news organisations explain their verification openly, trust follows naturally. Hiding methods breeds suspicion

Transparency is the cure for a lot of this. It costs effort but pays off. People notice honesty over time. Reputation becomes valuable again

[Serpent23]

Our family set up a code word after my aunt nearly sent money to someone pretending to be my cousin. The voice sounded exactly like him

She only hesitated because he called her by a nickname he never uses. Small details saved her

Now we have a word only we know. It feels a bit silly, but it works. Everyone should do it

Older relatives especially need this protection. Scammers target them deliberately