Content Moderation Case Study: Understanding Cultural Context To Detect Satire (2020)

from the she's-a-witch,-burn-her dept

Summary: During the somewhat controversial Senate confirmation hearings for the nomination of Judge Amy Coney Barrett to the Supreme Court, there were a few moments that gained extra attention, including a confrontation between Senator Mazie Hirono and the nominee concerning statements regarding LGBTQ rights that Barrett had made in the past. Hirono, who had separately called the hearings themselves illegitimate, was then criticized by traditionally right-leaning media for what they felt was overly aggressive questioning.

The satirical site The Babylon Bee, which frequently targets Democrats for satirization, published a piece roughly parodying a famous Monty Python sketch in which villagers in a medieval town try to determine if someone is a witch, including by weighing them to see if they weigh the same as a duck. The Babylon Bee took that sketch’s premise and ran a satirical article claiming that Hirono demanded that Barrett be weighed against a duck.

Facebook had the article removed, saying that it was “inciting violence.” The Babylon Bee appealed the decision, only to be told that upon a further “manual” review, Facebook had decided that its original analysis stood, and that the article “incites violence.”

Decisions to be made by Facebook:

• How do you handle moderation that requires understanding both current political controversies and historical cultural references?
• How do you distinguish actual satire from that which only pretends to be satire?
• How do you determine what is actually likely to incite violence?

Questions and policy implications to consider:

• How can rules against “inciting violence” be written to take into account satire and cultural references?
• Is it reasonable to expect that content moderators will understand cultural references as satirical?
• How much should a platform be expected to take into account the target audience of a particular website?

Resolution: As the tweet from The Babylon Bee’s CEO started to go viral, leading to another round of news coverage in traditionally right-wing focused publications, Facebook eventually apologized and said that the moderation decision (and the manual review) were a mistake.

"This was a mistake and we apologize that it happened. Satire can be difficult for our systems to identify, but we've restored the article and their ability to monetize," a Facebook spokesperson told Fox News.

As often happens in these situations, the CEO of the Babylon Bee insisted that this response was implausible, apparently believing that everyone would recognize the cultural references his site’s article was making use of for satire.

"Why did it have to take getting the media involved to fix this? And why did it happen in the first place?" Dillon asked in response to Facebook. "This was not just an algorithm flagging an article in error. Yes, that happened. But then a manual review took place and the ruling to penalize us was upheld. I notice they left that part out."

Originally posted to the Trust & Safety Foundation website.

Filed Under: amy coney barrett, content moderation, mazie hirono, monty python, satire
Companies: babylon bee, facebook

Content Moderation Case Study: Profanity Filter Causes Problems At Paleontology Conference (October 2020)

from the f-that dept

Summary: With the COVID pandemic still in full force, the Society of Vertebrate Paleontology moved its annual meeting online. The event was due to run for an entire week, but early issues caused attendees and moderators to question the contents of the pre-packaged content filter provided by Convey Services, which operated the virtual meeting software.

The effort to ensure followup Q&A sessions would be free of profanity and other disruptiveness went awry when terms commonly used by paleontologists got caught in the software's filter. While words like "pubic," "bone," or "hell" might be appropriately blocked elsewhere, the blocklisting of these words disrupted the conference the software was supposed to keep from being disrupted.

“I would hope that actual swears or slurs would be censored, since paleontology is not a field that's immune to racist/sexist jerks,” noted Brigid Christison, a masters' student in biology at Carleton University, in an email.

[...]

“Words like ‘bone,’ ‘pubic,’ and ‘stream’ are frankly ridiculous to ban in a field where we regularly find pubic bones in streams,” Christison said.

More problematically, attendees noted the software blocked "Wang" but left "Johnson" intact, despite both being sexual slang. This appeared to indicate some bias on the part of Convey's blocklist creators -- a bias that effectively erased a surname belonging to more than 90 million Chinese citizens.

Decisions to be made by Convey:

• Are generic ban lists effective enough in most situations to justify less content-specific alterations during more specialized use?
• Given the global reach of the internet, is it still acceptable to block common surnames that can also be deployed as sexual slang?
• Does providing separate keyword sets for different uses reduce the profitability of software sales and licenses?

Questions and policy implications to consider:

• Do word ban lists raise many issues in more general use or are problems noticed more often when the software is deployed by entities that deal with specialized subject matter?
• Does allowing licensees to alter keywords as needed eliminate some of the problems caused by non-specific ban lists?
• Is shifting the cost of moderation to customers a solid business decision?

Resolution: The Society was able to alter the ban list to better fit the subject matter once the problem was noticed. It continued to edit the keywords provided by Convey, reducing the probability of overblocking as the week rolled on. Because the alterations could be made on the client side, disruption was minimal, if inadvertently comical.

Originally posted to the Trust & Safety Foundation website.

Filed Under: blocklist, content moderation, paleontology, profanity filter

Content Moderation Case Study: Vimeo Moderates Uploads Of 'Commercial-Use' Videos Using Unclear Guidelines (2009)

from the what-is-commercial-use dept

Summary: Vimeo, the video-hosting website created by CollegeHumor's parent company in 2004, has always presented itself as a destination for creators who wished to free themselves from YouTube's limitations and aggressive monetization. Vimeo remains ad-free, supporting itself with subscription fees.

Other efforts were made to distance Vimeo from YouTube. Its fairly aggressive content policy forbade plenty of things that were acceptable on Google's platform, including videos promoting commercial services.

The terms of service didn't explicitly forbid content that related to commercial services but were not attempts to sell services directly to other Vimeo users, but user experience consultant Paul Boag found his videos targeted by Vimeo and given a week to move them to another hosting service. While some videos of Boag's rode the edge of the terms of service ban on commercial videos, others provided nothing more than marketing advice or reviews of browser plugins.

At that point, Vimeo also banned the embedding of hosted content on sites that also served up ads. Unfortunately for Boag, his own site contained ads, making it a violation of the terms of service to embed his own videos on his own site. And this rule wasn't set in stone: Vimeo rather unhelpfully clarified it did allow embedding on some sites with ads, but it was a decision only Vimeo could make.

Vimeo players cannot appear on domains running ads, its a decision we made in the beginning and have been going back and forth with allowing or disallowing it, but so far we cannot allow it unless it is with one of our partners.

Decisions to be made by Vimeo:

• Does distinguishing the service from YouTube with more onerous restrictions on content ultimately lower moderation costs by attracting a user base that self-selects?
• Is the risk of losing paid users an acceptable tradeoff for preventing Vimeo from "devolving" into just another YouTube-like service?
• Is making judgment calls on "commercial-use content" possible to do fairly when it appears to mainly be based on subjective calls by moderators?
• Is Vimeo large enough to comfortably absorb any damage to its reputation or user goodwill when its moderation decisions affect content that doesn't actually violate its policies?

Questions and policy implications to consider:

• Would allowing users to pay to upload commercial-use videos move the platform closer to the competitors Vimeo has tried to distance itself from?
• Would a transparent and open challenge process help Vimeo avoid losing paying users?

Resolution: Paul Boag's videos were removed and Boag chose to use a different platform to host his content, rather than continue to struggle with Vimeo's unclear content policies.

A few years later, Vimeo changed course and began allowing Pro users to upload content that was considered a violation of its terms of service in 2009. The restrictions on commercial-use content have since been rolled back even further, forbidding users from posting only certain kinds of commercial-use content:

We do not allow content that promotes:
 

• Illegal schemes (like Pyramid/Ponzi schemes)
• Businesses that promise wealth with little or no effort
• Unregistered securities offerings (absent a legal basis)
• Illegal products or services 
• Products or services (even if legal) using deceptive marketing practices 
• In addition, users may not use Vimeo.com’s messaging capabilities for unsolicited direct marketing purposes.

Originally posted on the Trust & Safety Foundation website.

Filed Under: commercial use, content moderation
Companies: vimeo

Content Moderation Case Study: Facebook's AI Continues To Struggle With Identifying Nudity (2020)

from the ai-is-not-the-answer dept

Summary: Since its inception, Facebook has attempted to be more "family-friendly" than other social media services. Its hardline stance on nudity, however, has often proved problematic, as its AI (and its human moderators) have flagged accounts for harmless images and/or failed to consider context when removing images or locking accounts.

The latest example of Facebook's AI failing to properly moderate nudity involves garden vegetables. A seed business in Newfoundland, Canada was notified its image of onions had been removed for violating the terms of service. Its picture of onions apparently set off the auto-moderation, which flagged the image for containing "products with overtly sexual positioning." A follow-up message noted the picture of a handful of onions in a wicker basket was "sexually suggestive."

Facebook's nudity policy has been inconsistent since its inception. Male breasts are treated differently than female breasts, resulting in some questionable decisions by the platform. Its policy has also caused problems for definitively non-sexual content, like photos and other content posted by breastfeeding groups and breast cancer awareness videos. In this case, the round shape and flesh tones of the onions appear to have tricked the AI into thinking garden vegetables were overtly sexual content, showing the AI still has a lot to learn about human anatomy and sexual positioning.

Decisions to be made by Facebook:

• Should more automated nudity/sexual content decisions be backstopped by human moderators?
• Is the possibility of over-blocking worth the reduction in labor costs?
• Is over-blocking preferable to under-blocking when it comes to moderating content?
• Is Facebook large enough to comfortably absorb any damage to its reputation or user goodwill when its moderation decisions affect content that doesn't actually violate its policies?
• Is it even possible for a platform of Facebook's size to accurately moderate content and/or provide better options for challenging content removals?

Questions and policy implications to consider:

• Is the handling of nudity in accordance with the United States' more historically Puritianical views really the best way to moderate content submitted by users all over the world?
• Would it be more useful to users is content were hidden -- but not deleted -- when it appears to violate Facebook's terms of service, allowing posters and readers to access the content if they choose to after being notified of its potential violation?
• Would a more transparent appeals process allow for quicker reversals of incorrect moderation decisions?

Resolution: The seed company's ad was reinstated shortly after Facebook moderators were informed of the mistake. A statement from the company raised at least one more question as its spokesperson did not clarify exactly what the AI thought the onions actually were, leaving users to speculate what the spokesperson meant, as well as how the AI would react to future posts it mistook for, "well, you know."

"We use automated technology to keep nudity off our apps," wrote Meg Sinclair, Facebook Canada's head of communications. "But sometimes it doesn't know a walla walla onion from a, well, you know. We restored the ad and are sorry for the business' trouble."

Originally posted at the Trust & Safety Foundation website.

Filed Under: ai, content moderation, nudity
Companies: facebook

Content Moderation Case Study: Scammers Targeting Scrabble Chat (2020)

from the it's-always-scrabble dept

Summary: In the spring of 2020, Mattel and Hasbro announced that the official mobile version of the game Scrabble would no longer be the game produced by Electronic Arts, but rather a new game called Scrabble Go created by a company called Scopely. The change drew the ire of fans (who have even started a petition for the old game to be brought back) for taking what had been a fairly standard mobile version of the popular word game, and introducing a new, flashier version that had some additional “gamification” incentives and put the focus on playing against others, rather than the computer as was typical in the previous game.

This also introduced a new feature: chat. Since players are playing against other human beings, Scopely decided to add a chat feature, but apparently did not consider how such features may be regularly abused. In the months since Scrabble Go launched, there have been many reports of so-called “romance scammers” trying to reach out to people via Scrabble Go’s chat feature.

Multiple reports of these kinds of approaches started appearing in various forums, with some examples of the scammers being quite persistent. At least in Australia, consumer protection officials noted that they have received multiple complaints of romance scammers approaching them via Scrabble Go. One woman in the UK noted that she has been approached by such scammers two to three times every week.

After three months of complaints, Scopely announced that it was rolling out an update that would allow players to “mute” the chat function.

Decisions to be made by Scopely:

• Does a mobile Scrabble game need a chat feature?
• If scammers are bothering players so often is the game better off without it?
• How will chat be monitored? Is there a program in place to catch and stop scammers?
• Are there other tools to limit the abuse of the chat feature?
• Should the default be that chat is open to all or should it be opt-in?

Questions and policy implications to consider:

• Any system that allows for person-to-person communication can be abused. How should companies looking to add useful features take this into account?
• How do you weigh the pros and cons of features like chat when comparing their usefulness for engagement against their trust-and-safety risks?

Resolution: A few months after launch, Scopely updated the app to allow players to mute the chat entirely. As complaints remained, it has also added an ability to only connect to friends you already know on Facebook or via your contacts (if you agree to upload your contacts to the service), effectively sandboxing the chat to only users the player has some connection with.

The company has also added the ability to “report” a chat if the user feels it is inappropriate.

Finally, to address the broader complaints about the game, Scopely introduced a “classic mode” to focus more on the traditional game, rather than all the bells and whistles of the full Scrabble Go.

Originally posted at the Trust & Safety Foundation website.

Filed Under: chat, content moderation, scammers, scams, scrabble

Content Moderation Case Study: Google's Photo App Tags Photos Of Black People As 'Gorillas' (2015)

from the automation-isn't-the-answer dept

Summary: In May 2015, Google rolled out its "Google Photos" service. This service allowed users to store their images in Google's cloud and share them with other users. Unlike some other services, Google's photo service provided unlimited storage for photos under a certain resolution, making it an attractive replacement for other paid services.

Unfortunately, it soon became apparent the rollout may have outpaced internal quality control. The built-in auto-tagging system utilizing Google's AI began tagging Black people as "gorillas," resulting in backlash from users and critics who believed Google's algorithm was racist.

Google's immediate response was to apologize to users. The Twitter user who first noticed the tagging error was contacted directly by Google, which began tackling the problem that made it out of beta unnoticed. Google's Yonatan Zunger pointed out the shortcomings of AI when auto-tagging photos, noting the company's previous problems with mis-tagging people (of all races) as dogs and struggles with less-than-ideal lighting or low picture resolution. In fact, Google's rollout misstep mirrored Flickr's own struggles with auto-tagging photos, which similarly resulted in Black people being labeled as "ape" or "animal."

Decisions to be made by Google:

• Would more diversity in product development/testing teams increase the chance issues like this might be caught before services go live?
• Can additional steps be taken to limit human biases from negatively affecting the auto-tag AI?
• Should more rigorous testing be performed in the future, given the known issues with algorithmic photo tagging?

Questions and policy implications to consider:

• Does seemingly inconsequential moderation like this still demand some oversight by human moderators?
• Will AI ever be able to surmount the inherent biases fed into it by those designing and training it?

Resolution: As of 2018, Google was still unable to completely eliminate this problem. Instead, it chose to eliminate the problematic tags themselves, resulting in no auto-tags for terms like "gorilla," "chimp," "chimpanzee," and "monkey." An investigation by Wired showed searches of Google Photos images returned zero results for these terms. Google said it was working on "longer-term fixes" but put no end date on when those fixes would arrive. It also acknowledged those terms had been blocked by Google and would remain blocked until the problem was solved.

Originally published at the Trust & Safety Foundation website.

Filed Under: ai, algorithmic bias, content moderation, google photos, racism
Companies: google

Content Moderation Case Study: Facebook Attracts International Attention When It Removes A Historic Vietnam War Photo Posted By The Editor-in-Chief Of Norway's Biggest Newspaper (2016)

from the the-terror-of-content-moderation dept

Summary: Tom Egeland, a Norwegian author of a number of best-selling fiction books, posted a well-known photo known as "The Terror of War" to Facebook. The historic photograph (taken by Vietnamese-American photographer Nick Ut) depicts a naked Vietnamese girl running from a napalm attack during the Vietnam War.

Ut's iconic photo brought the horrors of the war in Vietnam to viewers around the world. But it was not without controversy. Given the full-frontal nudity of the child depicted in the image, the Associated Press pushed back against Ut, citing the paper's policy against publishing nudity. In this case, the nudity of the child resulted in more resistance than usual. Ultimately, the AP decided to run the photo, resulting in a Pulitzer Prize for Ut in 1973.

Despite the photo's historical significance, Facebook decided to suspend Tom Egeland's account. It also deleted his post.

Facebook's decision was based on its terms of service. While the photo was undeniably a historical artifact, moderation efforts by the platform were not attuned to the history.

A notice sent to Egeland pointed out that any displayed genitalia would result in moderation. Also, given the platform's obligation to inform the government about Child Sexual Assault Material (CSAM), leaving a photo of a naked prepubscent up posed problems the algorithms couldn't necessarily handle on their own.

The decision to remove the post and suspend the author's account resulted in an open letter being sent by Norwegian journalist Epsen Hansen. The letter -- addressed to Facebook founder and CEO Mark Zuckerberg -- asked what negative effects moderation efforts like these would have on a "democratic society."

Decisions to be made by Facebook:

• Should automatic moderation that aids law enforcement be overridden when context shows posts are not attempting to sidestep rules put in place to prevent Facebook users from being subjected to abusive content?
• What value is placed on context-considerate moderation? Does it add or subtract from financial obligations to shareholders?
• Does it serve users better to be more responsive -- and helpful -- when context is a primary consideration?

Questions and policy implications to consider:

• Is the collateral damage of negative press like this offset by Facebook's willingness to be proactive when removing questionable content?
• Is it more important to serve private users than the numerous governments making moderation demands?
• Do inexact or seemingly-incoherent responses to controversial content raise the risk of government intervention?

Resolution: Despite the letter from a prominent Norwegian journalist, Facebook refused to reinstate the photo. Instead, it offered boilerplate stating its objection to "nude genitalia." While it stated it did make "allowances" for "educational, humorous, and satirical purposes." Ut's photo did not make the cut apparently. Facebook asked Aftenposten, Egeland, and/or Hansen to "pixelate" the iconic photo before reposting. This was the response from Aftenposten's Hegeland:

Unfortunately, Facebook did not see the pointed humor of Hansen's modification. Facebook's deletion of the original -- as well as its suspension of author Tom Egeland's account -- remained in force. While public shaming has had some effect on moderation efforts by social media companies, Facebook's stance on nudity -- especially the nudity of minors -- prevented it from backing down in the face of negative publicity.

Filed Under: content moderation, csam, historic photos, nudity, photo, tom egeland, vietnam
Companies: facebook

Content Moderation Case Study: GoFundMe Addresses Controversial Fundraising Efforts (2020)

from the you-are-the-company-you-keep dept

Summary: Shortly after protests began in Kenosha, Wisconsin over the shooting of Jacob Blake by police officers, armed citizens began showing up ostensibly to protect businesses and homes from violent protesters. One of these citizens was Kyle Rittenhouse, an Illinois native who traveled to Kenosha as a self-appointed peacekeeping force.

Following an altercation at a Kenosha car dealership, Rittenhouse shot three protesters, killing two of them. Shortly after it became apparent Rittenhouse was going to be criminally-charged, fundraisers for his legal defense were set up in his name at GoFundMe.

As controversy continued to swirl, GoFundMe deleted the fundraisers from its platform and refunded all donations. When asked for the reason, GoFundMe stated the fundraisers had violated its terms of service. While nothing was specifically cited by GoFundMe as the violation triggering the removals, its terms of service allow it to remove "any other activity" the site deems "unacceptable."

Decisions to be made by GoFundMe:

• Should moderation decisions be influenced by public outcry? Or should objectivity remain in place even if controversial fundraisers result in considerable public pushback?
• Does a policy of deleting fundraisers for people charged with serious criminal activity give users the impression that everyone accused of criminal acts is guilty? Does this prevent the wrongfully accused from securing competent legal help?
• Can fundraisers that violate the terms of service be detected prior to public posting? If this isn't possible at this time, would it be worth implementing to lower the possibility of controversial fundraisers gaining public attention before GoFundMe can address them?

Questions and policy implications to consider:

• Does a perceived lack of consistency in moderation decisions encourage more direct regulation by government agencies?
• Does GoFundMe have any obligation to treat accused criminals as innocent until proven guilty?
• Does treating the terms of service as fluid and subject to constant change deter fundraisers from using the site?

Resolution: GoFundMe removed the fundraisers linked to Kyle Rittenhouse shortly after they appeared. While its explanation to queries was non-specific, the site's terms of use allow GoFundMe almost unlimited discretion when it comes to hosting fundraising efforts.

This decision follows others GoFundMe has made in the past to take down controversial fundraising attempts. In 2015, it removed a fundraiser for Baltimore police officers facing criminal charges in the death of Freddie Gray. The stated reason for the removal was the fundraiser's attempt to "benefit" people charged with serious crimes. However, that same violation was not cited during the removal of Rittenhouse-related fundraisers, despite his being charged with first degree homicide.

Filed Under: content moderation, crowdfunding, funding, kyle rittenhouse
Companies: gofundme

Content Moderation Case Study: Using Fact Checkers To Create A Misogynist Meme (2019)

from the content-moderation-inception dept

Summary: Many social media sites include fact checkers in order to either block or at least highlight information that is determined to be false or misleading. However, in some ways, that alone can create a content moderation challenge.

Alan Kyle, a privacy and policy analyst, noticed this in late 2019 when he came across a meme picture on Instagram from a meme account called “memealpyro” showing what appeared to be a great white shark leaping majestically out of the ocean. When he spotted the image, it had been blurred, with a notice that it had been deemed to be “false information” after being “reviewed by independent fact checkers.” When he clicked through to unblur the image, next to the image there was a small line of text saying “women are funny.” And beneath that the fact checking flag: “See why fact checkers say this is false.”

The implication of someone coming across this image with this fact check is that the fact check is on the statement, leading to the ridiculous/misogynistic conclusion that women are not funny and that an independent fact checking organization had to flag a meme image suggesting otherwise.

As Kyle discusses, however, this seemed to be an attempt to rely on fact checkers checking one part of the content, in order to create the misogynistic meme. Others had been using the same image -- which was computer generated and not an actual photo -- and claiming that it was National Geographic’s “Picture of the Year.” This belief was so widespread that National Geographic had to debunk the claim (though it did so by releasing other, quite real, images of sharks to appease those looking for cool shark images).

The issue, then, was that fact checkers had been trained to debunk the use of the photo, on the assumption it was being posted with the false claim that it was National Geographic’s “Photo of the Year,” and Instagram’s system didn’t seem to expect that other, different claims might be appended to the same image. When Kyle clicked through to see the explanation, it was only about the “Picture of the Year” claim (which was not made on this image), and (obviously) not on the statement about women.

Kyle’s hypothesis is that Instagram’s algorithms were trained to flag the picture as false, and then possibly send the flagged image to a human reviewer -- who may have just missed that the text associated with this image was unrelated to the text for the fact check.

Decisions to be made by Instagram:

• If the caption and a picture need to be combined to be designated as false information, how should Instagram fact checkers handle cases where that information is separated?
• How should fact checkers handle mixed media content, in which text and graphics or video may be deliberately unrelated?
• Should automated tools be used to flag viral false information in a way that might be gamed?
• How much human review should be used for algorithmically flagged “false” information?

Questions and policy implications to consider:

• When there is an automated fact checking flagging algorithm, how will users with malicious intent try to game the system, as in the above example?
• Is fact checking the right approach to “meme’d” information that is misleading, but not in a meaningful way?
• Would requiring fact checking across social media lead to more “gaming” of the system as in the case above?

Resolution: As Kyle himself concludes, situations like this are somewhat inevitable, as the setup of content moderation works against those trying to accurately deal with content such as the piece described above:

There are many factors working against the moderator making the right decision. Facebook (Instagram’s parent company) outsources several thousand workers to sift through flagged content, much of it horrific. Workers, who moderate hundreds of posts per day, have little time to decide a post’s fate in light of frequently changing internal policies. On top of that, much of these outsourced workers are based in places like the Philippines and India, where they are less aware of the cultural context of what they are moderating.

The Instagram moderator may not have understood that it’s the image of the shark in connection to the claim that it won a NatGeo award that deserves the false information label.

The challenges of content moderation at scale are well documented, and this shark tale joins countless others in a sea of content moderation mishaps. Indeed, this case study reflects Instagram’s own challenged content moderation model: to move fast and moderate things. Even if it means moderating the wrong things.

Filed Under: content moderation, fact checking, memes
Companies: instagram

Content Moderation Case Study: Google Refuses A Law Enforcement Agency Demand To Remove A Video Depicting Police Brutality (2011)

from the government-takedowns dept

Summary: Google began documenting government requests for content removal in 2009. Periodic transparency reports informed users about demands made by government agencies, breaking requests down by country and targeted service (YouTube, Google search, Blogspot, etc.)

In October 2011, the section dealing with US government agency requests included a note indicating Google had refused a questionable demand to take down a YouTube video.

We received a request from a local law enforcement agency to remove YouTube videos of police brutality, which we did not remove. Separately, we received requests from a different local law enforcement agency for removal of videos allegedly defaming law enforcement officials. We did not comply with those requests, which we have categorized in this Report as defamation requests.

News coverage about the unusual request pointed out YouTube's value as an archive of public interest recordings. Later reports issued by Google added the name of the law enforcement agency: the Greensboro Police Department -- one that apparently has a history of officers deploying excessive force.

This isn't the only time Google has rejected an apparent effort to remove content that does not reflect well on the agency making the request. Subsequent reports show Google has rejected government requests targeting recordings of abuse of inmates by corrections officers, articles detailing a police officer's work history, videos containing information about a law enforcement investigation, and five requests to remove videos that "criticized local and state government agencies."

Decisions to be made by Google:

• Does targeted content contain sufficient public interest to justify rejecting government demands?
• Does pushing back on requests usually result in government agencies withdrawing or dropping questionable takedown demands?
• If copyright claims are raised as justification for seemingly inappropriate requests, is fair use raised in defense of leaving the video up?

Questions and policy implications to consider:

• Does YouTube have a duty to preserve content with sufficient public interest, even if the original uploader/creator tries to remove it?
• Does Google/YouTube have a duty to the public in general, even though it's a private company? If so, does this conflict with Google's obligations to its shareholders?
• Does challenging questionable government demands provide more value to users?

Resolution: As noted in Google's transparency report, the targeted video was not taken down. This initial oddity was followed by other questionable requests over the years. Following this showdown with a local law enforcement agency, Google began highlighting requests that "may be of public interest," allowing users to gain more insight into questionable government activities, as well as similarly-notable requests made by private citizens and businesses.

Filed Under: content moderation, government requests, law enforcement, police brutality, takedowns
Companies: google, youtube