Falsified and Altered Media Policy

Overview

Sharing synthetic, manipulated, or out-of-context media that could deceive or mislead people and potentially cause harm is prohibited here ("misleading media").

There is no labeling feature on Goodbye App. We cannot mark a post as disputed, attach context to it, or reduce how far it travels. The only two things this policy can do are leave a post alone or remove it. We also do not screen anything before it is posted, and we run no software that looks for altered media, so a report is how this reaches us. One person runs Goodbye App and reads every report.

What is in violation of this policy?

To adhere to our policy concerning misleading media, encompassing images, videos, audios, gifs, and URLs hosting relevant content, all of the following conditions are necessary for removal:

1. The media should be likely to create widespread confusion on public matters, impact public safety, or cause serious harm.
2. The media should exhibit significant and deceptive alteration, manipulation, or fabrication.
3. The media must be presented as a true record of something real, with false context around it.

All three have to be true at once. Media that meets one or two of them is left alone.

When a report brings misleading media to us, these are the criteria we apply:

1. Has the content been significantly and deceptively altered, manipulated, or fabricated?

For content to be subject to removal under this policy, we must be reasonably convinced that the media have undergone significant and deceptive alterations, manipulations, or fabrications. Synthetic and manipulated media can take various forms, utilizing a wide array of technologies. Key factors we consider include:

• Whether media depicting real individuals have been fabricated or simulated, particularly through the utilization of artificial intelligence algorithms.
• Whether there are any additions, edits, or removals of visual or auditory information (like new video frames, overdubbed audio, or modified subtitles) that substantially alter the understanding, meaning, or context of the media.
• Whether the media have been extensively edited or post-processed in a manner that fundamentally changes their composition, sequence, timing, or framing, leading to a distortion of their meaning.
• Whether the media have been enhanced or filtered during creation, editing, or post-processing in a way that fundamentally alters the understanding, meaning, or context of the content.

We do not remove media that have undergone non-significant edits, such as retouched photos or color-corrected videos. We have no detection technology of our own, and no arrangement with any third party that checks media for us. What we have is the report, and what a person can see in the post it names.

Where we cannot reliably tell whether media have been altered or fabricated, we leave them alone.

2. Is the content presented deceptively or with misleading context?

We also read the words around the media, because those are on the page and a person reading a report can see them. What we are looking for is whether the post states something about the media that is not true. Some of the contextual elements we read:

• Whether the context surrounding the identity of the individuals or entities depicted in the media is false or misleading.
• Whether the media are accompanied by false or misleading information regarding their source, location, time, or authenticity.
• Whether fictional, inauthentic, or staged media are portrayed as real or endorsed as factual events, including reenactments or productions presented as genuine occurrences.
• Whether the media are shared with misrepresentations or false quotes, presenting fabricated claims as factual depictions.

We do not remove media accompanied by commentary or opinions that do not mislead or present a false context, as outlined above. Where we cannot reliably tell whether media have been shared with false context, we leave the content alone.

3. Content featuring misleading media is subject to removal under this policy if it poses a risk of causing serious harm.

We evaluate specific harms such as:

• Threats to privacy or hindrance of free expression and civic participation, including stalking, harassment, intimidation, or voter suppression
• Encouragement of abusive behavior towards individuals or groups
• Potential for mass violence or widespread civil unrest
• Interference with public services, protection efforts, or emergency response
• Physical threats to individuals or groups

We also assess the timeframe in which the content could impact public safety or cause serious harm. Immediate harm is a decisive factor for removal under this policy.

Content containing misleading media that does not pose an immediate threat, but still carries the potential to impact public safety, cause harm, or create confusion on public issues like health, environment, safety, human rights, immigration, and social-political stability, is not removed. There is nothing in between here: we cannot label it or reduce how far it travels, so it stays up.

Other rules here address the same harms from a different side: violent threats, civic integrity, and hateful conduct. Where a post breaks one of those, it is dealt with under that policy, whether or not the media in it have been altered.

What Doesn't Violate This Policy?

This policy is about media that have been altered or falsified. It is not about subjects that are difficult or contested. In the absence of other policy breaches, the following do not contravene this policy:

1. Commentary, reviews, opinions, or reactions shared alongside media edits that solely add commentary, reviews, opinions, or reactions. This facilitates further debate and discourse on various topics and is not considered a violation of this policy.
2. Animations, illustrations, and cartoons, provided they do not significantly confuse viewers about the authenticity of the media.
3. Counterspeech. A direct response to misleading information that corrects the record, or that points to accurate information, is not itself misleading media.
4. Memes or satire, as long as they do not significantly confuse viewers about the authenticity of the media.

A memory misremembered is not this policy's concern either. If you post an old photograph and give the wrong year, the wrong place, or the wrong name, that is not falsified media, and nothing here is aimed at it. This policy is about media that have been altered or fabricated, not about a person getting a detail of their own life wrong.

Pictures of your own people are not what this policy is for. A torn photograph repaired, a faded one colorized, a blurred face sharpened, a portrait built by software from the few pictures that survive: none of that breaks this policy. Neither does a picture of a person who has died, made so that a child who never met them has something to look at. All three conditions above have to be true at once, and the third one is the one these fail, because a picture given to your family as what it is has no false claim around it.

If you want to be sure, say in the post what was done to the picture. You do not have to. It is worth doing anyway, because the people you are writing to may look at it for a long time after you are gone, and they will want to know what in it is the person and what is the software.

Consequences of Policy Violations:

For severe violations of this policy, particularly involving misleading media that pose a significant risk of harm to individuals or communities, we mandate the removal of such content.

In certain instances, immediate and permanent account suspension may be implemented for policy violations. For most breaches, temporary account lockouts may precede further actions.

Persistent breaches, after a warning, may result in permanent account suspension. Every one of those decisions is made by a person reading the account, not by an automated system, and you are told when it happens, in the app and at the email address on the account.

If you see misleading media here, report it. That is the only way it reaches us, and a person reads every report.