Posts Tagged: Responsibility
Anonymous claims responsibility for Moscow traffic jam tied to app exploit
On Thursday morning, Moscow’s busy Fili district became the site of a traffic jam unlike any before it. Motherboard (via The Verge) reports hackers used Russia’s Yandex Taxi ride-hailing app to order dozens of drivers to coverage on Kutuzovsky Prospekt, one of the city’s main thoroughfares. The act caused traffic on part of the already congested street to come to a standstill for about 40 minutes while Yandex worked to address the situation.
“On the morning of September 1st, Yandex Taxi encountered an attempt by attackers to disrupt the service — several dozen drivers received bulk orders to the Fili district of Moscow,” a Yandex spokesperson told Motherboard. In a separate statement shared with Russia’s state-owned TASS news agency, Yandex said it reworked its routing algorithm following the attack to prevent similar incidents from occurring in the future. The event is one of the first known instances of hackers exploiting a ride-hailing app to create a traffic jam.
Someone hacked #YandexTaxi and ordered all available taxis to Kutuzov Prospect in Moscow
Now there is a huge traffic jam with taxis.
It‘s like James Bond movie. pic.twitter.com/IatuAEtA2i
— Russian Market (@runews) September 1, 2022
Several Twitter accounts claiming affiliation with Anonymous say the hacktivist collective is behind the incident. On Friday, one Anonymous account said the group worked with the IT Army of Ukraine, a volunteer organization formed at the start of the war, to carry out the attack.
Anonymous previously claimed responsibility for a cyberattack that took down multiple Russian government websites, including those belonging to the Kremlin and the Ministry of Defence. “Faced with this series of attacks that Ukraine has been suffering from the Russian dictator Vladimir Putin, we could not help but support the Ukrainian people,” the group said at the time.
The Four Rs of Responsibility, Part 2: Raising authoritative content and reducing borderline content and harmful misinformation
Our community guidelines set the rules of the road on YouTube, and a combination of people and machines help us remove more violative content than ever before. That said, there will always be content on YouTube that brushes up against our policies, but doesn’t quite cross the line. So over the past couple of years, we’ve been working to raise authoritative voices on YouTube and reduce the spread of borderline content and harmful misinformation. And we are already seeing great progress. Authoritative news is thriving on our site. And since January 2019, we’ve launched over 30 different changes to reduce recommendations of borderline content and harmful misinformation. The result is a 70% average drop in watch time of this content coming from non-subscribed recommendations in the U.S.1
Raising authoritative voices on YouTube
More and more people turn to YouTube to catch up on the latest news or simply learn more about the topics they’re curious about — whether it’s climate change or a natural disaster. For topics like music or entertainment, relevance, newness and popularity are most helpful to understand what people are interested in. But for subjects such as news, science and historical events, where accuracy and authoritativeness are key, the quality of information and context matter most — much more than engagement. That’s why we’ve re-doubled our efforts to raise authoritative sources to the top and introduced a suite of features to tackle this challenge holistically:
- Elevating authoritative sources in our systems: In 2017, we started to prioritize authoritative voices, including news sources like CNN, Fox News, Jovem Pan, India Today and the Guardian, for news and information queries in search results and “watch next” panels. Let’s say you’re looking to learn more about a newsworthy event. For example, try searching for “Brexit.” While there will be slight variations, on average, 93% of the videos in global top 10 results come from high-authority channels. Authoritativeness is also important for evergreen topics prone to misinformation, such as videos about vaccines. In these cases, we aim to surface videos from experts, like public health institutions, in search results. Millions of search queries are getting this treatment today and we’re continually expanding to more topics and countries.
- Providing reliable information faster for breaking news: Reliable information becomes especially critical as news is breaking. But as events are unfolding, it can take time to produce high-quality videos containing verified facts. So we’ve started providing short previews of text-based news articles in search results on YouTube, along with a reminder that breaking and developing news can rapidly change. We’ve also introduced Top News and Breaking News sections to highlight quality journalism. In fact, this year alone, we’ve seen that consumption on authoritative news partners’ channels has grown by 60 percent.
- Providing context to users: Sometimes a video alone does not provide enough context to viewers about what they are watching. We want to make sure that people who watch videos about topics prone to misinformation are provided additional information while viewing. To that end, we’ve designed a variety of information panels that target different types of context, such as general topics and recent news prone to misinformation, or about publishers themselves. For example, when people watch videos that encourage viewers to skip the MMR vaccine, we show information panels to provide more basic scientific context, linking to third-party sources. Or if people are viewing news videos uploaded by a public broadcaster or a government-funded news outlet, we show informational notices underneath the video about the news outlet. Collectively, we’ve delivered more than 3.5 billion impressions across all of these information panels since June 2018 and we’re expanding these panels to more and more countries.
Reducing borderline content and harmful misinformation
Content that comes close to — but doesn’t quite cross the line of — violating our Community Guidelines is a fraction of 1% of what’s watched on YouTube in the U.S. To give a quick comparison, meditation videos (a fairly narrow category) have more daily watch time than borderline and harmful misinformation combined. That said, even a fraction of a percent is too much. So this past January, we announced we’d begin reducing recommendations of borderline content or videos that could misinform users in harmful ways. This work is still ramping up and we’ve expanded to more countries outside of the U.S., including the UK, Ireland, South Africa and other English-language markets. And we have begun expanding this effort to non-English-language markets, starting with Brazil, France, Germany, Mexico and Spain.
So how does this actually work? Determining what is harmful misinformation or borderline is tricky, especially for the wide variety of videos that are on YouTube. We rely on external evaluators located around the world to provide critical input on the quality of a video. And these evaluators use public guidelines to guide their work. Each evaluated video receives up to 9 different opinions and some critical areas require certified experts. For example, medical doctors provide guidance on the validity of videos about specific medical treatments to limit the spread of medical misinformation. Based on the consensus input from the evaluators, we use well-tested machine learning systems to build models. These models help review hundreds of thousands of hours of videos every day in order to find and limit the spread of borderline content. And over time, the accuracy of these systems will continue to improve.
Our work continues. We are exploring options to bring in external researchers to study our systems and we will continue to invest in more teams and new features. Nothing is more important to us than ensuring we are living up to our responsibility. We remain focused on maintaining that delicate balance which allows diverse voices to flourish on YouTube — including those that others will disagree with — while also protecting viewers, creators and the wider ecosystem from harmful content.
[Read more] The Four Rs of Responsibility, Part 1: Removing harmful content
From the timeline:
July 27, 2015; https://youtube.googleblog.com/2015/07/youtube-comments.html
Sept 14, 2016; https://www.blog.google/outreach-initiatives/google-news-initiative/digital-news-initiative-introducing/
July 20, 2017; https://youtube.googleblog.com/2017/07/bringing-new-redirect-method-features.html
Feb 2, 2018; https://youtube.googleblog.com/2018/02/greater-transparency-for-users-around.html
July 9, 2018; https://youtube.googleblog.com/2018/07/building-better-news-experience-on.html
July 9, 2018; https://youtube.googleblog.com/2018/07/building-better-news-experience-on.html
July 9, 2018; https://youtube.googleblog.com/2018/07/building-better-news-experience-on.html
March 7, 2019; https://india.googleblog.com/2019/04/bringing-greater-transparency-and.html
June 3, 2019; https://youtube.googleblog.com/2019/06/an-update-on-our-efforts-to-protect.html
June 5, 2019; https://youtube.googleblog.com/2019/06/our-ongoing-work-to-tackle-hate.html
July 8, 2019; https://youtube-creators.googleblog.com/2019/08/preserving-openness-through-responsibility.html
The Four Rs of Responsibility, Part 1: Removing harmful content
Over the next several months, we’ll provide more detail on the work supporting each of these principles. This first installment will focus on “Remove.” We’ve been removing harmful content since YouTube started, but our investment in this work has accelerated in recent years. Below is a snapshot of our most notable improvements since 2016. Because of this ongoing work, over the last 18 months we’ve reduced views on videos that are later removed for violating our policies by 80%, and we’re continuously working to reduce this number further.1
Developing policies for a global platform
Before we do the work of removing content that violates our policies, we have to make sure the line between what we remove and what we allow is drawn in the right place — with a goal of preserving free expression, while also protecting and promoting a vibrant community. To that end, we have a dedicated policy development team that systematically reviews all of our policies to ensure that they are current, keep our community safe, and do not stifle YouTube’s openness.
After reviewing a policy, we often discover that fundamental changes aren’t needed, but still uncover areas that are vague or confusing to the community. As a result, many updates are actually clarifications to our existing guidelines. For example, earlier this year we provided more detail about when we consider a “challenge” to be too dangerous for YouTube. Since 2018, we’ve made dozens of updates to our enforcement guidelines, many of them minor clarifications but some more substantive.
For particularly complex issues, we may spend several months developing a new policy. During this time we consult outside experts and YouTube creators to understand how our current policy is falling short, and consider regional differences to make sure proposed changes can be applied fairly around the world.
Our hate speech update represented one such fundamental shift in our policies. We spent months carefully developing the policy and working with our teams to create the necessary trainings and tools required to enforce it. The policy was launched in early June, and as our teams review and remove more content in line with the new policy, our machine detection will improve in tandem. Though it can take months for us to ramp up enforcement of a new policy, the profound impact of our hate speech policy update is already evident in the data released in this quarter’s Community Guidelines Enforcement Report:
The spikes in removal numbers are in part due to the removal of older comments, videos and channels that were previously permitted. In April 2019, we announced that we are also working to update our harassment policy, including creator-on-creator harassment. We’ll share our progress on this work in the coming months.
Using machines to flag bad content
Once we’ve defined a policy, we rely on a combination of people and technology to flag content for our review teams. We sometimes use hashes (or “digital fingerprints”) to catch copies of known violative content before they are ever made available to view. For some content, like child sexual abuse images (CSAI) and terrorist recruitment videos, we contribute to shared industry databases of hashes to increase the volume of content our machines can catch at upload.
In 2017, we expanded our use of machine learning technology to help detect potentially violative content and send it for human review. Machine learning is well-suited to detect patterns, which helps us to find content similar (but not exactly the same) to other content we’ve already removed, even before it’s ever viewed. These systems are particularly effective at flagging content that often looks the same — such as spam or adult content. Machines also can help to flag hate speech and other violative content, but these categories are highly dependent on context and highlight the importance of human review to make nuanced decisions. Still, over 87% of the 9 million videos we removed in the second quarter of 2019 were first flagged by our automated systems.
We’re investing significantly in these automated detection systems, and our engineering teams continue to update and improve them month by month. For example, an update to our spam detection systems in the second quarter of 2019 lead to a more than 50% increase in the number of channels we terminated for violating our spam policies.
Removing content before it’s widely viewed
We go to great lengths to make sure content that breaks our rules isn’t widely viewed, or even viewed at all, before it’s removed. As noted above, improvements in our automated flagging systems have helped us detect and review content even before it’s flagged by our community, and consequently more than 80% of those auto-flagged videos were removed before they received a single view in the second quarter of 2019.
We also recognize that the best way to quickly remove content is to anticipate problems before they emerge. In January of 2018 we launched our Intelligence Desk, a team that monitors the news, social media and user reports in order to detect new trends surrounding inappropriate content, and works to make sure our teams are prepared to address them before they can become a larger issue.
We’re determined to continue reducing exposure to videos that violate our policies. That’s why, across Google, we’ve tasked over 10,000 people with detecting, reviewing, and removing content that violates our guidelines.
For example, the nearly 30,000 videos we removed for hate speech over the last month generated just 3% of the views that knitting videos did over the same time period.
Last week we updated our Community Guidelines Enforcement Report, a quarterly report that provides additional insight into the amount of content we remove from YouTube, why it was removed, and how it was first detected. That report demonstrates how technology deployed over the last several years has helped us to remove harmful content from YouTube more quickly than ever before. It also highlights how human expertise is still a critical component of our enforcement efforts, as we work to develop thoughtful policies, review content with care, and responsibly deploy our machine learning technology.
2 Nov 16, 2016; https://youtube.googleblog.com/2016/11/more-parental-controls-available-in.html
2 June 18, 2017; https://www.blog.google/around-the-globe/google-europe/four-steps-were-taking-today-fight-online-terror/
2 July 31, 2017; https://youtube.googleblog.com/2017/07/global-internet-forum-to-counter.html
2 Aug 1, 2017; https://youtube.googleblog.com/2017/08/an-update-on-our-commitment-to-fight.html
2 Dec 4, 2017; https://youtube.googleblog.com/2017/12/expanding-our-work-against-abuse-of-our.html
2 April 23, 2018; https://youtube.googleblog.com/2018/04/more-information-faster-removals-more.html
2 Dec 1, 2018; https://youtube.googleblog.com/2019/06/an-update-on-our-efforts-to-protect.html
2 Jan 15, 2019; https://support.google.com/youtube/thread/1063296?hl=en
2 Feb 19, 2019; https://youtube-creators.googleblog.com/2019/02/making-our-strikes-system-clear-and.html
2 Feb 28, 2019; https://youtube-creators.googleblog.com/2019/02/more-updates-on-our-actions-related-to.html
2 June 5, 2019; https://youtube.googleblog.com/2019/06/our-ongoing-work-to-tackle-hate.html
2 July 1, 2019; https://support.google.com/youtube/thread/8830320
2 Aug 21, 2019; https://support.google.com/youtube/thread/12506319?hl=en
2 Coming soon; https://youtube.googleblog.com/2019/06/taking-harder-look-at-harassment.html