A Facebook test that promoted comments containing the word fake to the top of news feeds has been criticised by users.
The trial, which Facebook says has now concluded, aimed to prioritise “comments that indicate disbelief”.
It meant feeds from the BBC, the Economist, the New York Times and the Guardian all began with a comment mentioning the word fake.
The test, which was visible only to some users, left many frustrated.
The comments appeared on a wide range of stories, from ones that could be fake to ones that were clearly legitimate. The remarks, which would appear at the top of the comments section, came from a variety of people but the one thing that they had in common was the word fake.
“Clearly Facebook is under enormous pressure to tackle the problem of fake news, but to question the veracity of every single story is preposterous,” said Jen Roberts, a freelance PR consultant.
“Quite the reverse of combating misinformation online, it is compounding the issue by blurring the lines between what is real and what isn’t. My Facebook feed has become like some awful Orwellian doublethink experiment.”
In a statement, Facebook told the BBC: “We’re always working on ways to curb the spread of misinformation on our platform, and sometimes run tests to find new ways to do this. This was a small test which has now concluded.
“We wanted to see if prioritising comments that indicate disbelief would help. We’re going to keep working to find new ways to help our community make more informed decisions about what they read and share.”
Facebook has been under enormous pressure to deal with the issue of fake news since it was singled out as one of the main distribution points for hoax stories during the US presidential election.
Twitter plans to increase the number of characters in tweets from 140 to 280 for the majority of users.
The new limit will not apply to tweets written in Japanese, Chinese and Korean which can convey more information in a single character.
The move follows a trial among a small group of users which started in September in response to criticism that it was not easy enough to tweet.
The change is part of Twitter’s plan to attract new users and increase growth.
During the test, only 5% of tweets sent were longer than 140 characters and only 2% more than 190, the social media site said in a blog post.
But those who did use the longer tweets, got more followers, more engagement and spent more time on the site, it added.
“During the first few days of the test, many people tweeted the full 280 limit because it was new and novel, but soon after behaviour normalised,” wrote Aliza Rosen, Twitter’s product manager.
“We saw when people needed to use more than 140 characters, they tweeted more easily and more often. But importantly, people tweeted below 140 most of the time and the brevity of Twitter remained.”
According to Twitter, 9% of tweets in English hit the character limits.
“This reflects the challenge of fitting a thought into a tweet, often resulting in lots of time spent editing and even at times abandoning tweets before sending,” Ms Rosen said.
Increasing the character limit should not affect people’s experience on the site, she added.
“We – and many of you – were concerned that timelines may fill up with 280-character tweets, and people with the new limit would always use up the whole space. But that didn’t happen.”
When the change was announced, many criticised it, pointing out changes they would rather see, such as a crackdown on hate crime and bots, and the introduction of a chronological timeline and edit function.
The site currently has 330 million active users. This compares with 800 million for Instagram and more than 2 billion users for Facebook.
Facebook is testing a system that allows users to message themselves their nude photos in an effort to combat so-called revenge porn.
It will store a “fingerprint” of images to prevent any copies of them being shared by disgruntled ex-lovers.
The trial is in Australia, where studies suggest one in five women aged 18-45 may have had image-based abuse.
But one expert says there will still be problems outside Facebook and related sites such as WhatsApp and Instagram.
Facebook said it looked forward “to getting feedback and learning” from the trial.
Revenge porn is a growing issue in Australia, according to e-safety commissioner Julie Inman Grant, who is working with Facebook on the trial.
“We see many scenarios where maybe photos or videos were taken consensually at one point, but there was not any sort of consent to send the images or videos more broadly,” she told ABC News.
She sought to reassure potential victims who might be concerned about proactively sending themselves intimate photos.
“It would be like sending yourself your image in email, but obviously this is a much safer, secure end-to-end way of sending the image without sending it through the ether,” she said.
“They’re not storing the image, they’re storing the link and using artificial intelligence and other photo-matching technologies.”
Users wanting to take part in the trial must first file a report with the commissioner, who will in turn share it with Facebook.
Prof Clare McGlynn, an expert from Durham Law School, described it as “an innovative experiment”.
“I welcome Facebook taking steps to tackle this issue, as it has often been very slow to act in the past. However, this approach is only ever going to work for a few people and when we think of the vast number of nudes taken and shared each day, this clearly isn’t a solution,” she told the BBC.
In March, Facebook was embroiled in a scandal when it emerged that a 30,000-strong private members group, Marine United, was routinely sharing images of nude women.
The group – made up of US marines – shared photographs of naked and semi-naked female colleagues.
In response to the revelations, Facebook introduced a feature that tagged pictures reported to it as revenge porn using photo-matching technology. It used this to prevent the image spreading and closed down the majority of accounts reported to it as hosting such images.
The world’s most profitable firm has a secretive new structure that would enable it to continue avoiding billions in taxes, the Paradise Papers show.
They reveal how Apple sidestepped a 2013 crackdown on its controversial Irish tax practices by actively shopping around for a tax haven.
It then moved the firm holding most of its untaxed offshore cash, now $252bn, to the Channel Island of Jersey.
Apple said the new structure had not lowered its taxes.
It said it remained the world’s largest taxpayer, paying about $35bn (£26bn) in corporation tax over the past three years, that it had followed the law and its changes “did not reduce our tax payments in any country”.
Up until 2014, the tech company had been exploiting a loophole in tax laws in the US and the Republic of Ireland known as the “double Irish”.
This allowed Apple to funnel all its sales outside of the Americas – currently about 55% of its revenue – through Irish subsidiaries that were effectively stateless for taxation purposes, and so incurred hardly any tax.
Instead of paying Irish corporation tax of 12.5%, or the US rate of 35%, Apple’s avoidance structure helped it reduce its tax rate on profits outside of the US to the extent that its foreign tax payments rarely amounted to more than 5% of its foreign profits, and in some years dipped below 2%.
The European Commission calculated the rate of tax for one of Apple’s Irish companies for one year had been just 0.005%.
Apple came under pressure in 2013 in the US Senate, when CEO Tim Cook was forced to defend its tax system.
Angry that the US was missing out on a huge amount of tax, then-Senator Carl Levin told him: “You shifted that golden goose to Ireland. You shifted it to three companies that do not pay taxes in Ireland. These are the crown jewels of Apple Inc. Folks, it’s not right.”
Mr Cook responded defiantly: “We pay all the taxes we owe, every single dollar. We do not depend on tax gimmicks… We do not stash money on some Caribbean island.”
After the EU announced in 2013 that it was investigating Apple’s Irish arrangement, the Irish government decided that firms incorporated there could no longer be stateless for tax purposes.
In order to keep its tax rates low, Apple needed to find an offshore financial centre that would serve as the tax residency for its Irish subsidiaries.
In March 2014, Apple’s legal advisers sent a questionnaire to Appleby, a leading offshore finance law firm and source of much of the Paradise Papers leak.
It asked what benefits different offshore jurisdictions – the British Virgin Islands, Bermuda, the Cayman Islands, Mauritius, the Isle of Man, Jersey and Guernsey – could offer Apple.
The document asked key questions such as was it possible to “obtain an official assurance of tax exemption” and could it be confirmed that an Irish company might “conduct management activities… without being subject to taxation in your jurisdiction”.
They also asked whether a change of government was likely, what information would be visible to the public and how easy it would be to exit the jurisdiction.
Source document: Apple questionnaire (extract)
Leaked emails also make it clear that Apple wanted to keep the move secret.
One email sent between senior partners at Appleby says: “For those of you who are not aware, Apple [officials] are extremely sensitive concerning publicity. They also expect the work that is being done for them only to be discussed amongst personnel who need to know.”
Apple chose Jersey, a UK Crown dependency that makes its own tax laws and which has a 0% corporate tax rate for foreign companies.
Paradise Papers documents show Apple’s two key Irish subsidiaries, Apple Operations International (AOI), believed to hold most of Apple’s massive $252bn overseas cash hoard, and Apple Sales International (ASI), were managed from Appleby’s office in Jersey from the start of 2015 until early 2016.
This would have enabled Apple to continue avoiding billions in tax around the world.
Apple’s 2017 accounts showed they made $44.7bn outside the US and paid just $1.65bn in taxes to foreign governments, a rate of around 3.7%. That is less than a sixth of the average rate of corporation tax in the world.
Apple’s Tim Cook calls the EC ruling “total political crap”, with “no reason for it in fact or in law”. Ireland says the EU is encroaching on sovereign taxation. It fears multinationals will go elsewhere.
Ireland agrees to collect the €13bn, to be held in a managed escrow account pending the appeal verdict.
In October 2017, the EU says it will take Ireland to court as it has not yet collected the money. Ireland says it is complicated and it needs time.
When the “double-Irish” loophole was shut down, Ireland also created new tax regulations that companies like Apple could take advantage of.
One of the companies that Apple moved to Jersey, ASI, had rights to some of Apple Inc’s hugely valuable intellectual property.
If ASI sold the intellectual property back to an Irish company, the Irish company would be able to offset the enormous cost against any future profits. And since the IP holder, ASI, was registered in Jersey, the profits of the sale would not be taxed.
It appears Apple has done just that. There was an extraordinary 26% spike in Ireland’s GDP in 2015 which media reports put down to intellectual property assets moving into Ireland. Intangible assets rose a massive €250bn in Ireland that year.
Ireland’s department of finance denied that the new regulations had been brought in to benefit multinationals.
It said Ireland was “not unique in allowing companies to claim capital allowances on intangible assets” and had followed “the international norm”.
Apple declined to answer questions about its two subsidiaries moving their tax residency to Jersey.
It also declined to comment when asked whether one of those companies had helped create a huge tax write-off by selling intellectual property.
Apple said: “When Ireland changed its tax laws in 2015, we complied by changing the residency of our Irish subsidiaries and we informed Ireland, the European Commission and the United States.
“The changes we made did not reduce our tax payments in any country. In fact, our payments to Ireland increased significantly and over the last three years we’ve paid $1.5bn in tax there.”
The papers are a huge batch of leaked documents mostly from offshore law firm Appleby, along with corporate registries in 19 tax jurisdictions, which reveal the financial dealings of politicians, celebrities, corporate giants and business leaders.
Computers may be able to beat humans at Chess and Go but they are a poor second when it comes to video game StarCraft.
One of the world’s best human players trounced AI-controlled opponents 4-0 in a StarCraft tournament this week.
Social media giant Facebook was behind one of the robotic opponents that took on South Korean StarCraft ace Song Byung-gu.
The complexity of video games has led many AI researchers to create programs that can play them.
Sejong University in Seoul hosted the StarCraft tournament and this year was the first in which the AI-controlled programs, known as bots, played against a human opponent. Before now the bots only played each other.
There are several other tournaments based around StarCraft and bots and most just let the computers play each other. One did feature a top StarCraft player who also beat every bot.
StarCraft is a game involving a three-way conflict between humans, the insectoid Zerg and the strange Protoss aliens. Playing involves building a base to gather resources that help make combat units that are used to seek out and destroy opponents.
The complexity of StarCraft and the demands it places on players to plan and adapt strategy as well as act quickly have drawn attention from AI researchers.
As well as Facebook, bots for the StarCraft tournament were made by AI researchers in Norway, Australia and South Korea.
The bots did not pose much of a challenge for Mr Song, who took a total of 27 minutes to defeat all four. The longest match lasted over 10 minutes and the shortest just four and a half, reports the Technology Review.
Mr Song overcame the bots even though they could act much more swiftly than any human could. Top StarCraft players can make up to 600 actions per minute as they gather resources, explore the game map and fight opponents.
By contrast, the Norwegian bot in the contest carried out 19,000 actions per minute.
After the matches were over, Mr Song said some of the ways the bots reacted to his actions were “stunning” but said their overly cautious strategy was their undoing.
AI-controlled opponents look set to get more challenging, as earlier this year Blizzard released tools that let people make bots that can play StarCraft II. The research team behind Google’s DeepMind have declared their intention to take this game on.
Other groups have taken on other games. One bot created by the non-profit OpenAI research group has managed to defeat top players of Steam’s Dota 2 arena battle game. Its capabilities were shown off during the global tournament for Dota 2, The International, over the summer.
Also, a team at the Massachusetts Institute of Technology has created a bot that plays Super Smash Bros, Melee.
Tech companies need to do more to combat the “exponential” growth in child sexual exploitation online, Home Secretary Amber Rudd has said.
Ms Rudd said internet giants have a “moral duty” to act and need to work with smaller platforms used by child gamers where paedophiles operate.
She is to meet counterparts in the US government to discuss the issue.
The tech companies have said they are doing their utmost to keep their young users safe.
During Ms Rudd’s trip she will attend a roundtable discussion joined by tech companies including Google, Facebook and Microsoft.
The home secretary will welcome work that has been done to tackle online child sexual abuse, but will also say that more needs to be done at a “far greater pace” across the technology industry.
Rudd to techies: Stop patronising us
Message encryption a problem – Rudd
Speaking to the BBC’s Andrew Marr on Sunday, Ms Rudd said: “We’ve seen the real growth of child sexual exploitation internationally, and we’re going to make sure that we work with the Americans to take action against it.
“We need to make sure [tech companies] put their technical know how into addressing it.
“Particularly working with smaller platforms where children go to game online, to meet each other; there are paedophiles working there.
“We need to make sure that internet companies work with us, in partnership, to change this.”
Ms Rudd’s comments come as new government figures show there was a 700% increase in the number of indecent images identified on technology company servers and flagged to law enforcement agencies between 2013 and 2017.
Each month there are more than 400 arrests for indecent images of children offences in the UK and some 500 children are being protected from online sexual exploitation, the government said.
Ms Rudd will also raise concerns about the use of messaging websites and apps, as well as video and image apps and websites, during her trip.