Image copyrightEPA

A leaked email from an employee warned Uber’s bosses that its self-drive cars were involved in multiple crashes, only days before one killed a pedestrian.

Robbie Miller, operations manager for Uber’s self-driving trucks, wrote to the firm’s top executives saying the cars were “routinely in accidents”.

This was partly down to faults with the technology and partly because of the “poor behaviour” of operators, he said.

Uber has yet to respond to the revelations.

However, it told news and information website The Information in a statement: “Right now the entire team is focused on safely and responsibly returning to the road in self-driving mode.

“We have every confidence in the work that the team is doing to get us there.

“Our team remains committed to implementing key safety improvements, and we intend to resume on-the-road self-driving testing only when these improvements have been implemented and we have received authorisation from the Pennsylvania Department of Transportation.”

• Why you have (probably) already bought your last car
• Uber ends Arizona driverless car programme
• Toyota suspends US driverless car tests after fatal Uber accident

Mr Miller’s email was sent to the head of Uber’s autonomous vehicle unit, Eric Meyhofer, six other executives and lawyers.

Several of the drivers were not “properly vetted or trained”, he wrote.

It was sent on 13 March, only days before a fatal collision in Tempe, Arizona in which an autonomous Uber car hit and killed pedestrian Elaine Herzberg.

Uber suspended all its tests following the accident, which is still under investigation by the National Transportation Safety Board.

In June it was revealed that the safety operator of the car was watching TV just before the accident.

In his email, Mr Miller wrote: “The cars are routinely in accidents resulting in damage. This is usually the result of poor behaviour of the operator or the AV (autonomous vehicle) technology.

“A car was damaged nearly every other day in February. We shouldn’t be hitting things every 15,000 miles.”

He went on to describe three incidents, including one where a car “drove on the sidewalk for several metres”. In another, an accident was avoided only because a vehicle in the adjacent lane swerved to miss the Uber prototype, according to his email.

Mr Miller complained that such incidents were “ignored”, with the logs left unreviewed by senior managers for days or even weeks.

“This is not how we should be operating,” he said.

He recommended that Uber reduce its fleet size by up to 85%, stop tests after all accidents and review any incident, even minor ones “immediately”.

Mr. Miller is currently working at a self-driving truck start-up led by Anthony Levandowski, who was accused of stealing secrets from Google’s self-driving division Waymo.

Mr Levandowski was fired by Uber after he declined to testify in the lawsuit filed by Waymo against the firm. The case was settled in February with Uber giving a stake of its company to Waymo and agreeing not to use its technology in its self-driving cars.

Mr. Miller also worked at Mr Levandowski’s prior company Otto, which was acquired by Uber, and worked for him at Waymo.

Image copyrightHagen Hopkins/Getty Images

New Zealand’s justice minister has said Google could face prosecution for sending a mass email which included the name of the man accused of murdering British backpacker Grace Millane.

The suspect was granted a temporary name suppression while he awaits trial, making it illegal to publish his name.

The email, which named the suspect prominently, went to people signed up to receive the country’s top trends.

Google said it did not know about the suppression order.

Justice Minister Andrew Little told the New Zealand Herald that if the email breach was traced to any of Google’s New Zealand infrastructure then the firm could be prosecuted in the country. Google has an office in New Zealand with at least 20 staff and was set to add more employees this year.

“Google has staff in New Zealand, I know because I’ve got Christmas cards from them,” Mr Little said.”They should not be allowed to say ‘it’s all the machine’s problem it’s nothing to do with us.’ The truth is, Google is responsible for publishing in New Zealand information that’s been suppressed by a court.

“They’ve acted in contempt of court accordingly. We have to find a way of calling those folks to account,” TVNZ quoted him as saying.

Google did not respond to a request for comment from the BBC. But a spokesperson told The Herald that initial investigations by the company showed it did not know about the suppression order.

“When we receive valid court orders, including suppression orders, we review and respond appropriately. In this case, we didn’t receive an order to take action,” a statement quoted by New Zealand media says.

Several British news outlets have also named the murder suspect.

100,000 searches

In New Zealand’s court system, the accused and victims can ask to have their name suppressed, which means it becomes illegal to publish it in newspapers, online or anywhere else.

The purpose is to protect people not yet proven guilty but also to have a fairer trial by ensuring the jury is not prejudiced by media coverage.

• ‘Solo backpacking didn’t kill Grace’

In the Grace Millane case the accused was not given a suppression order by the Auckland district court, but his defence said they would appeal to a higher court – which automatically triggers a 20-working day suppression.

New Zealand media have upheld the suppression order, but some UK media have named the accused.

Mr Little rebuked them for this, after which they geo-blocked their stories so that they cannot be seen in New Zealand.

But then Google sent out an email to the subscribers of its top trends list with the name of the accused as well as articles linking to stories about the murder case. The email was first seen by New Zealand website The Spinoff.

It said more than 100,000 searches were made of the man’s name in the country.

The incident has sparked a debate in the country about suppression laws in the internet age.

When asked about this Prime Minister Jacinda Ardern said: “There is no doubt the environment has changed and that’s having an effect on a number of our laws in this space – reporting around suicide, you could even argue clean slate legislation [which helps people conceal their criminal record]. At this time it’s not part of our work.”

Image copyrightVevo

YouTube Rewind 2018, the annual review of the platform’s highlights, has become its most “disliked” video.

It has almost 10.1 million “dislikes”, with more than 82% of its viewers saying they did not enjoy it.

With almost 9.9 million “dislikes”, the video for Justin Bieber’s 2010 song Baby had filled the bottom slot for several years.

And fans’ comments below Baby reveal many “disliked” Rewind in a deliberate effort to bump Bieber off the bottom.

“Justin must be feeling so relieved now,” posted one YouTube member, whose comment rapidly attracted more than 1,000 likes itself.

‘Giant ad’

The Rewind video has been widely criticised for failing to feature some of the site’s top performers, such as Logan Paul and PewDiePie, and for ignoring significant web-culture events.

PewDiePie, also known as Felix Kjellberg, joined the chorus of criticism in a video saying Rewind featured too much of the Fortnite game and too few YouTubers.

Prominent YouTuber Marques Brownlee said Rewind had been a “big celebration” of YouTubers and the biggest events that had happened on the site in a particular year.

“It became an honour to be included in Rewind,” said Mr Brownlee.

But now, he said, YouTube saw Rewind as a way to “showcase all the best stuff that happens on YouTube for advertisers”.

Instead of honouring creators, said Mr Brownlee, it was now a list of “advertiser-friendly” content.

“Rewind has turned into a giant ad for YouTube,” he said.

Image copyrightFacebook

The mother of a stillborn child has called on tech companies to rethink how they target ads after she was inundated with baby-related promotions.

Gillian Brockell wrote to Facebook, Instagram, Twitter and Experian, saying if they were smart enough to deduce she had been pregnant, they should have realised her baby had died.

Other internet users have remarked that they have had similar experiences.

Facebook and Twitter have acknowledged they could do better.

‘Heartbroken’

Washington DC-based Ms Brockell posted a message to Twitter last month to share the news that her son had died in the womb.

She suggested that the technology companies should have picked up on this or other online activity resulting from her son’s death.

Instead, she said, the companies remained focused on her earlier pregnancy-related posts and actions.

“Did you not see the three days of silence, uncommon for a high-frequency user like me?” she wrote.

“And then the announcement with keywords like ‘heartbroken’ and ‘problem’ and ‘stillborn’ and the 200 teardrop emoticons from my friends? Is that not something you could track?”

She added that when she had tried to actively discourage the technology companies from showing her the pregnancy-related promotions, they had misinterpreted her response.

“When we… click ‘I don’t want to see this ad,’ and even answer your ‘Why?’ with the cruel-but-true ‘It’s not relevant to me,’ do you know what your algorithm decides?

“It decides you’ve given birth, assumes a happy result, and deluges you with ads for the best nursing bras… tricks to get the baby to sleep through the night… and the best strollers to grow with your baby.

“And then, after all that, Experian swoops in with the lowest tracking blow of them all: a spam email encouraging me to ‘finish registering your baby’ (I never ‘started’ but sure) to track his credit throughout the life he will never lead.”

‘Working on it’

Facebook’s advertising chief, Rob Goldman, has been the first of the executives responsible to respond.

He apologised for Ms Brockell’s experience but noted that the platform’s settings included an option to block ads about topics the user might find painful, including parenting.

“It still needs improvement, but please know that we’re working on it,” he added.

Ms Brockell thanked him for his reply but said the solution was not ideal.

She later added that even after she had taken the suggested measure, she was still shown an advert that suggested she consider adoption, which was inappropriate given the circumstances.

Other users have also reported that Facebook’s Hide Ad Topics setting does not always have the desired effect.

Last month, an English mother of a stillborn girl wrote an open letter to the social network after she told it to hide parenting ads as well as anything related to babies, family and the home.

She said she had still been targeted with baby-related products.

“Your ads were unintentionally taunting me with reminders of what I’d lost,” she wrote.

Media playback is unsupported on your device

Facebook told the BBC this had been caused by a bug in its system that had since been fixed.

Twitter has also issued a brief statement.

“We cannot imagine the pain of those who who have experienced this type of loss,” it said.

“We are continuously working on improving our advertising products to ensure they serve appropriate content to the people who use our services.”

The BBC understands Experian plans to comment later after it has spoken directly to Ms Brockell.

Image copyrightMBK Media

A robot on show at a Russian state-sponsored event has turned out to be a man dressed in a costume.

Robot Boris featured on Russian TV and was apparently able to walk, talk and dance.

But soon after its appearance journalists began to question the bot’s authenticity.

In a picture published afterwards on social media, the neck of a person was clearly visible.

The robot is in fact a 250,000 rouble (£2,975) costume called Alyosha the Robot, made by a company called Show Robots.

• Robot dons gown in virtual graduation ceremony
• ‘Robot surgeon’ helps Northern Ireland cancer patients

While the organisers of the Proyektoria technology forum – which was aimed at youngsters – had not claimed the robot was real, the TV coverage on Russia-24 suggested it was.

Russian website TJournal raised concerns about the robot asking a series of questions:

• why did the robot feature no sensors?
• how did Russian scientists get the robot made so quickly, with no papers published about it beforehand?
• why there has been no internet coverage of such an advanced robot?
• why did robot make so many unnecessary movements during its dance?
• why did it look like a man would fit perfectly inside it?
• why offer a pre-recording of its voice rather than do it live?

On the website of the firm behind the Alyosha robot costume the product is described as being able to create “an almost complete illusion that you have a real robot”.

Image copyrightReuters

A row has broken out in Ghana over the government’s decision to buy unmanned drones to deliver blood and essential medicines.

The government wants the drones to deliver to hospitals nationwide, especially in remote areas.

Some MPs say they want more spent on clinics and ambulances instead of what they call a high-tech vanity project.

Government minister Pius Enam Hadzide rejected the criticism, saying the plan sought to improve health services.

• BBC Africa Live: More on this and other stories

The Ghana Medical Association called for the immediate suspension of the newly approved drones deal, arguing that it would not solve the country’s health problems and needed to be tested on a trial basis first.

Mr Hadzide, the deputy minister of information, said he was “taken aback” by the comments, saying the association had been consulted. But he also said “our doors remain open for further engagement and consultation”.

The scheme was recently approved along partisan lines by 102 votes to 58, and is set to be implemented next year.

Media playback is unsupported on your device

A US company, Zipline International Inc, is to be paid more than $12m (£9.5m) to run the project for four years.

Opposition MP Cassiel Ato Forson said the government had negotiated an “extremely bad” deal with Zipline.

The BBC’s Favour Nunoo in Ghana says that earlier this year it was reported that the country had fewer than 55 ambulances serving a population of 29 million.

He says this has angered some members of the public who think the drones scheme is not appropriate given the challenges that exist in the health sector.

The case of drone use in Rwanda has been cited by supporters of the Ghana deal, our correspondent says, but critics argue that Rwanda at the time had a more robust healthcare system.

Image copyrightReuters

A report from the US Congress has revealed that credit agency Equifax’s 2017 network breach, which affected 143 million people, was not spotted because of an expired software certificate

Last week, mobile operator O2 blamed a similar issue for causing a network blackout which affected the UK.

But what is a digital certificate and why do they expire?

And will similar administrative errors continue to dog the industry and cause widespread havoc?

What is a digital certificate?

Digital certificates are basically small pieces of code created by using sophisticated mathematics that ensure that communication between devices or websites are sent in an encrypted manner, and are therefore secure.

They play an essential role in keeping IT infrastructure up and running safely and are issued by certificate authorities, who electronically vouch that the certificates are genuine. When issued, these certificates are given an expiration date of anything between a few months and several years.

Digital certificates are issued for a variety of software that encrypts communications, including those embedded in hardware.

In O2’s case it seems that a certificate linked to network equipment installed by Ericsson was the weak link.

• O2 4G data network restored after day-long outage
• Equifax says almost 400,000 Britons hit in data breach

Equifax’s certificate was linked to crucial software that monitored the network for suspicious traffic, meaning the hackers were not spotted in time.

While some think that the reason they expire is to allow the authorities to keep charging for renewals, there are some valid reasons why they need to be regularly updated – including changing technology, new vulnerabilities to encryption and the ownership of the certificate changing hands.

What went wrong?

In O2’s case, the certificate reached its expiry date, which in turn meant that when different parts of the network attempted to communicate securely, they no longer trusted each other and refused to connect.

The details about what caused O2’s network to fail have not yet been made public but commentators are speculating.

“So, imagine it was a web server certificate that failed. Suddenly it would have tried to make a secure connection with another piece of equipment which would have replied, ‘no, I can’t trust you’ and rejected it,” said Prof Alan Woodward, a computer scientist from University of Surrey.

“Some of this equipment is 10 years old and the programmer may have put in a certificate with a 10-year shelf life, thinking ‘this will last’.”

In the worst-case scenario, someone would have to physically go to the affected equipment, whether it be a web server or a phone mast, to put a new certificate on it.

“I can’t imagine how many bits of equipment needed a manual update,” said Prof Woodward.

In Equifax’s case, the certificate in question was linked to software which monitored the network for suspicious traffic and had expired 19 months ahead of the breach.

“That means that they weren’t monitoring their network for hackers for a long time and I think they will come in for a lot of criticism for that,” said Prof Woodward.

Are there likely to be more failures?

There are billions of certificates in circulation and, with the internet of things flourishing and connecting ever more devices to the web, more are needed each day.

What is needed is a mechanism to make sure they are renewed when necessary, said Tim Callan, a senior fellow at certificate issuer Sectigo.

“As business becomes digital in increasingly complex and ubiquitous ways, all enterprises need to protect themselves from repeating this disastrous outcome. A best practice in so doing is to automate the discovery, monitoring, and renewal of certificates of all types.

“The proliferation of certificates and ever-increasing complexity of IT infrastructure has made it more and more challenging for IT professionals to stay on top of this component of their networks.”

Image copyright Eyevine/New York Times

The woman who created and sold what many recognise as the world’s first word processor has died aged 93.

Evelyn Berezin called the device the Data Secretary when, in 1971, her company Redactron launched the product.

She grew Redactron from nine employees to close to 500 and was named one of the US’s top leaders by BusinessWeek magazine in the year she sold it, 1976.

She had earlier built one of the original computerised airline reservation systems.

The innovation – which matched customers and available seats – was tested by United Airlines in 1962.

According to the Computer History Museum, it had a one-second response time and worked for 11 years without any central system failures.

The technology vied with the rival Sabre system, developed by American Airlines, for being the first of its kind.

In addition, Ms Berezin helped pioneer other types of special-purpose computing including:

• an automated banking system
• a weapons-targeting calculator for the US Defense Department
• terminals for a horse-racing track that monitored how much money was being bet on each animal

Screenless editing

In an interview in 2015, Ms Berezin explained that she had decided to set up her own business in the mid-60s after coming to the conclusion that her prospects were limited so long as she was employed by someone else.

She said that she had initially considered developing an electronic cash register but ultimately opted to create what would become known as a word processor instead.

She said: “6% of all the people in the United States who worked were secretaries.

“At the time we started, which was in 1968 to 1969, nobody really had any desk-type computers on which you could write a word-processing program that a secretary would use.

“I know that desktop computers seem obvious now but it wasn’t so then.”

At the time, the nearest equivalent was a machine by IBM called the MT/ST – a typewriter with magnetic tape recording and playback facilities.

IBM’s marketing referred to a “word processor”, but the machine relied on relay switches rather than computer chips, had been targeted at military equipment makers rather than the wider business market, and in Ms Berezin’s mind was “klutzy”.

“We were committed to building a computer to run our system and we knew that we had to use integrated circuits because it was the only way we could make it small enough and cheap enough and reliable enough to sell,” Ms Berezin said.

Her machine – which stood about 1m (3ft 3in) tall – featured a keyboard, cassette drives, control electronics and a printer.

It could record and play back what the user had typed, allowing it to be edited or reprinted.

The original model lacked a monitor, and soon faced competition from a rival, the Lexitron, which did.

But later versions of the Data Secretary did include a screen.

Sparks and water

The project nearly ended in disaster.

Ms Berezin had intended to buy the processors required from Intel, which had gone into business in 1968. But it said it was too busy dealing with orders for its memory chips.

The solution was that Redactron had to design some of the chips required itself and provide the schematics to two manufacturers.

There were further problems with a prototype when it was put on display in a New York hotel for reporters to see.

The issue was that in dry weather, it was prone to a build-up of static electricity, which caused sparks to fly between its circuits, preventing it from working.

“To our horror it was a dry day and the engineers were setting this non-working machine up for our big story,” Ms Berezin said.

“Ed Wolf [our head of engineering brought] a full pail of water and without a word to anyone throws the pail of water over the whole thick carpet in the room.

“The water sank into the carpet, which stayed damp for three or four hours, and the machine worked perfectly.”

The first production machine was delivered to a customer in September 1971. And over the following year, Redactron sold or rented more than 770 others, excluding demo units.

Over the following years, demand grew but the company’s finances came under strain, in part because of high interest rates and a recession that meant clients wanted to rent rather than buy its products.

“We were told by the bank to sell the company and they had somebody they knew who was interested,” said Ms Berezin.

“At the time, I was distraught about it.”

She went to work for the purchaser, the business equipment-maker Burroughs Corporation. But it proved to be an ill match.

“I was not one of them – I told them what I thought – a loud woman they did not know how to deal with,” she said.

“So, they disconnected and so did I.”

Ms Berezin left the company around 1980, after which she became involved in venture capital and sat on other companies’ boards before becoming involved with Stony Brook University.

The New York Times reported that a nephew had confirmed she had died on 8 December in Manhattan after turning down treatment for cancer.

One of the remaining Data Secretary word processors can be seen on display at the Computer History Museum in California.

Image copyrightGetty Images

Twitter boss Jack Dorsey has responded to criticism of his 10-day meditation retreat in Myanmar, saying: “I don’t know enough and need to learn more.”

Last year, Myanmar’s military launched a violent crackdown after Rohingya militants attacked police posts.

Since then thousands of people have been killed and over 700,000 Rohingya refugees have fled the country.

Mr Dorsey said he was “aware of the human rights atrocities” in Myanmar, also known as Burma.

But he added that the trip had been “purely personal” and had focused only on his meditation practice.

Mr Dorsey said he had been asked by Twitter users to explain what the social media company was doing about the crisis in Myanmar.

He tweeted Twitter was a way for people to “bear witness to the plight of the Rohingya and other peoples and communities”.

“We’re actively working to address emerging issues,” he said.

Birthday retreat

Mr Dorsey had previously posted tweets about a vipassana meditation retreat he had taken to celebrate his birthday.

Vipassana is a Buddhist meditation method said to provide practitioners with greater insight into their inner selves.

Describing Myanmar as an “absolutely beautiful country”, Mr Dorsey had encouraged any of his four million followers who were interested and willing to travel there and try it for themselves.

The UN has described Myanmar’s operation against the Rohingya as a “textbook example of ethnic cleansing” and says senior officials should be investigated and tried for genocide.

The army has previously cleared itself of all wrongdoing and rejects the UN’s allegations.

Criticism of Mr Dorsey’s choice of holiday destination came in quickly to the social media platform.

“Writing what is effectively a free tourism advert for them at this time is reprehensible,” one Twitter user wrote in response to Mr Dorsey’s tweets.

“The tone-deafness here is… wow,” another user said.

“This is an extremely irresponsible recommendation,” one response reads. “Does he pay no attention to the news and the outcry on his own platform?” they added.

Others have responded positively, with one Twitter user writing: “Glad you got to experience Myanmar – it’s an incredible place with even more incredible people.”

But another wrote: “Dude – just stop.

“You’re just making it worse.”