RPA (Robotics Process Automation) — using robots to automate work processes —has been touted as a possible way to solve the issue of labour shortages and, at the same time, increase productivity by transforming the way we work.

It has been introduced in finance and other fields, producing great results in automating document creation and data entry tasks.

Nevertheless, many companies still need to carry out tasks such as recording minutes of meetings and transcribing speeches. While AI and software that automatically converts speech to text are already available on the market, converting speech to text accurately still needs to be done manually.

How can we solve this issue and help create a society that is easy to work in?

Toshiba provides an answer with its newly-developed speech recognition AI.

In an interview, Taira Ashikawa, head of Research and Hiroshi Fujimura, lead researcher of the Toshiba Corporate R&D Centre’s Media AI Laboratory, which developed AI, talked about the history of speech recognition using AI and the breakthroughs they had made during development.

Smooth speech transcription with a fast, easily readable display: Toshiba has a history of working on media intelligence, a field which makes use of human voices and images that have undergone information processing. The foundation the company has cultivated in the field over many years plays a big role in the creation of this voice recognition AI.

Toshiba first began developing AI in 2015. At the time, there was increasing momentum around the world in the field of information accessibility, which aims to create environments that enable people that are deaf and hard of hearing to access and input information. Toshiba has started “Universal Design (UD) Adviser System” since 2007 to enable employees with disabilities to participate in product development. The company believes in promoting diversity and inclusion in the workplace and develops UD-friendly products and services through the years.

“When we interviewed the hearing-impaired people in UD Adviser System, we found out that they wanted to participate in meetings and lectures in real time, and not just read the transcripts provided subsequently. So we tried to provide a function that would automatically display easy-to-read subtitles in real time. To assist the hearing-impaired people in collecting and providing information, we need to do two things: expand information accessibility for the hearing-impaired, and increase productivity. The development of speech recognition AI started from these two points in mind.” Ashikawa said.

The technology behind the accuracy in speech recognition: When you describe speech from people’s conversations during meetings and lectures, you will end up with a text that is hard to read. Anyone who has ever transcribed speeches can tell you that. There is a lot of unnecessary content that gets in the way of getting information such as meaningless filler words like “Uh,” and “Umm” and expressions of agreement that add nothing to the content.

The speech recognition AI Toshiba developed is able to recognise speech with high accuracy and detect fillers and hesitation markers as well. This is an essential function when it comes to increasing productivity. Algorithms form the core of AI, and the development team explored a variety of approaches to increase accuracy.

“At first we hit a wall because the level of accuracy of recognition just wouldn’t increase no matter what we did. Our primary goal was to provide users with something that they could use conveniently. By using the increasingly popular model known as LSTM (*1) as well as CTC learning (*2), we tried to teach AI about speech peculiarities such as fillers and hesitation markers that are exclusive to human beings.” Fujimura said.

(*1) LSTM (Long Short-term Memory): one of the developed forms of RNN (Recurrent Neural Network), which has a recursive structure in a hidden layer. It is able to learn long-term dependency relationships which are difficult for conventional RNNs to do.

(*2) CTC (Connectionist Temporal Classification): A method for training RNN to solve problems where sequence lengths differ during input by introducing null characters and adjusting loss functions.

Up until now, speech recognition has worked by analysing sound wave patterns and parsing them by identifying that this part is “a,” this other part is “i” and so on. However, fillers and hesitation markers have an endless variety of patterns, and it would take a long time to learn about them one by one.

“We used LSTM to capture information such as ‘this is what fillers are like,’ ‘this is what it sounds like when someone hesitates over a word,’ as a statistical model and then used CTC learning to make the AI learn it as a model. Through that, the AI became capable of detecting the countless patterns of fillers and hesitation markers as well,” Fujimura said.

“There is still plenty of room for improvement in development and technology to achieve a fully accurate speech recognition offering. Our speech recognition AI can recognize speech in Japanese, English and Chinese for now. We strive to develop an environment where speakers of different languages will be able to enjoy a smooth conversation with one another. When we develop AI, we dream of taking something like that, which you only see in futuristic science fiction or comic books, and making it a reality”.

This is how the AI evolved into speech recognition AI with superior accuracy. When the development team used lectures as an opportunity for verification testing, the AI achieved an average speech recognition ratio of 85%. That means it was able to recognize the contents of speech above a certain level without editing or advance learning. Now that they have raised the accuracy of the speech recognition, they are considering applying it to the communication AI known as RECAIUS™.

They developed applications where a representative affair is a real-time subtitle display function for the hearing-impaired people. They harness AI to display speech clearly with fillers and hesitation markers reflected in faint, non-obtrusive subtitles. This was a user-friendly specification introduced following detailed discussions with users.

“As far as we’re concerned, filler words like “umm” and “uhh” just get in the way. However, what the hearing-impaired people really want is to get as much information as possible. When they read the subtitles while following the movements of the speaker’s lips, they get stressed when fillers and hesitation markers are cut out because they feel that the speaker is saying something that isn’t being reflected in the text,” Ashikawa said.

“So we decided to leave the fillers and hesitation markers in the subtitles but is displayed faintly to make the text easier to read. However, when we record them as transcribed documents, we remove the filler and hesitation markers. That way, we get brief and concise documents.”

AI shows its true worth in manufacturing as well: In March 2019, Toshiba collaborated with Dwango Co Ltd and held a live broadcast of the 81st National Convention of the Information Processing Society of Japan on video website “niconico”. Subtitled videos were distributed online in real time. They are planning to deploy it not only for office tasks but for use in manufacturing settings as well.

“It’s rare to see speech recognition being used as a service in offices today. So it would be ideal for us if users would trust our product and use it, and if it could become something they used in everyday business without being conscious that it’s a speech recognition AI. For example, the words we are speaking right now could become a text polished enough to be used as a business document, with the speakers clearly identified to show who said what. We hope to create a speech recognition AI that is handy and reliable,” Ashikawa said.

“The use of speech recognition hasn’t been applied in manufacturing sites. However there is a need for hands-free voice collection and recording in factories during maintenance and inspections. So I think there’s room for this speech recognition AI to be adopted there as well,” Fujimura said.

We hope to use our knowledge and know-how about manufacturing facilities to seamlessly integrate speech recognition into their operations. We can do that because we have spent a long time developing speech recognition AI and accumulated knowledge about manufacturing and infrastructure settings. ’Why does Toshiba work on speech recognition?’ I think this will provide one of the answers to that fundamental question”.

With the numerous potential applications and benefits, there is no doubt that this speech recognition software will be making its presence increasingly felt in more offices and manufacturing sites in the near future.

Each fall, she pens a letter to her 11th grade English class at T.C. Williams High School International Academy in northern Virginia. She tells the students who she is: where she’s from, the jobs she has held, which TV shows she favors. Then, she asks for a reply.

“I read all of those letters,” Reamer said. Over the next few weeks, “I think about it. I come up with three to five books for each kid, and we sit down, face-to-face, to read the jackets.”

She picks the possibilities from a meticulously curated library of almost 1,000 books she houses on shelves painted turquoise and burnt-orange in her third-floor classroom – a library she paid for through online fundraisers and grants. Reamer, 45, offers the teens texts meant to feel familiar: The characters might resemble her students, practice their religion, speak their language.

Reamer knows she has found the right book, she said, “when the kid just lights up – there should be an ‘Ooooooh!’ noise.”

She is seeing early success, which she attributes both to the personalized selections and to her habit of reading aloud to students. In the four years since Reamer began teaching at the Alexandria, Virginia, campus, her students – all immigrant or international students with limited English – have on average achieved two years’ worth of reading progress each year.

Reamer is a rare bright spot at a moment when children’s reading proficiency is plunging nationwide. A study released last year by the National Center for Education Statistics found that two-thirds of fourth- and eighth-graders in the United States do not meet basic federal standards for reading proficiency, capping a decade’s worth of poor performance.

In Virginia, students’ reading scores on the National Assessment of Educational Progress – an exam widely known as “the nation’s report card” – have declined since 2017.

Bob Farrace, spokesman for the National Association of Secondary School Principals, said lackluster literacy levels contribute to a larger disaffection with learning. He pointed to a 2016 Gallup survey that found just one-third of 11th- and 12th-graders nationally report feeling “engaged” by what they study in school. Although Reamer is teaching to students who come from abroad, Farrace said, her methods offer a path for educators everywhere.

“She’s clearly empowering kids to take control of their reading and thus, their learning,” Farrace said. “Allowing for that kind of choice needs to become a schoolwide culture.”

Farrace said it’s troubling – if typical – that Reamer had to raise money to make it possible. Federal data show that nine in 10 educators spend almost $500 each year on school supplies. The Washington Post reported that teachers often go to extreme lengths to find classroom resources, begging friends for help, visiting Goodwill or scouring garage sales.

Reamer estimates she garnered roughly $10,000 over the past three years by soliciting donations on the website DonorsChoose, while $5,000 more came from grants and from an Alexandria parent-teacher group.

This month, she’s hosting two DonorsChoose fundraisers: one for paperbacks and magazines “that should fit in any teen’s back pockets” and one for “a selection of classics.”

It takes time – hours every week – to stay on top of her grant applications and fundraisers, Reamer said. She often works on both late into the night, a level of devotion she acknowledged may be impossible for some teachers.

“I don’t mind sitting on the couch with a cocktail and writing a grant, but if you have kids you might not have time,” she said.

For the most part, Reamer said, she feels well supported by Alexandria City Public Schools. T.C. Williams boasts a well-stocked library, she said, staffed by knowledgeable librarians – and she knows that is not the case for many schools.

“In some places, entire school libraries are in the situation where they have to do fundraising to receive books,” said Audrey Church, director of the School Librarianship graduate program at Longwood University.

Reamer started her first fundraiser three years ago because she wanted to build a book collection tailored to her students, one boasting texts at every reading level. She also hoped to collect stories featuring diverse characters.

“When a student named Meena from Afghanistan can come in, and I can hand her a book with a main character named ‘Meena,’ ” Reamer said, “you can bet she’s going to read that whole book.”

Nationally, the push for diverse books gained serious momentum about a half-decade ago, Church said, with the founding of the nonprofit We Need Diverse Books. The movement is built around a three-part theory, Church said: that texts should serve as mirrors, windows and sliding glass doors.

Children must see themselves in what they read – that’s the mirror piece, Church said. But they should also read about people who are different – the window. And they must “make that connection that we’re all different, but we’re all alike, and we should be accepting of all,” Church said. That’s the sliding door.

Over the past two years, Alexandria City Public Schools – where 72% of students are black, Hispanic or Asian – has moved to diversify its library and textbook offerings. In 2018, the school system set aside $1.2 million to purchase more than 70,000 “diverse and culturally sensitive” books and related materials for students in kindergarten through fifth grade.

In the fall, Alexandria City schools won $60,000 from the nonprofit First Book, which gives books to disadvantaged children nationwide. That money will allow more than 10,000 children to take home between one and three books over the next few months, said Shanna Samson, the administrator who applied for the grant.

At each school in Alexandria, educators will work to select books – often, bilingual texts – that match the demographics of their classrooms.

“As our nation becomes more and more diverse, we just need to be more representative of our students,” Samson said. “Our content must be representative of our country.”

At the start of the year, Reamer used a small percentage of the money she has received to purchase furniture for Reamer’s Reading Retreat, set up in a cozy corner.

Nearly every morning, students arrive before school starts, grab a book and settle down in one of three canvas chairs – and a rocking chair – to snatch a blissful 15 minutes cocooned among colorful pillows and patterned blankets. Reamer estimates roughly three-fourths of her current and former students read books on their own, in their free time.

Spaced around Reamer’s classroom, at least four separate signs bear the same verb: “READ.”

Shekofa Hussaini, 18, needs no instruction.

Hussaini, a student in Reamer’s Honors English class, packs the book she’s reading – David Levithan’s “Some Day” – into her school bag each morning. Even if she doesn’t find time to open it, Hussaini said, she likes to have the novel nearby.

It’s the second installment in a four-part saga centered on “A,” the protagonist who is cursed to awaken in a different body every morning. Reamer recommended the series after Hussaini – who emigrated from Afghanistan three years ago – confided she likes to “get to know a lot of different people.”

The books resonated for another reason.

In childhood, after watching a medical drama on television, Hussaini decided she wanted to be a doctor. In Afghanistan, that felt impossible. Even after moving to the United States, it still seems daunting.

But then she thinks about A’s never-failing determination to court love interest Rhiannon – no matter the protagonist’s inability to remain in one body.

A’s “perseverance reminds me of my own,” Hussaini said. “I read, and know that I will not give up.”

Assistant Attorney General John Demers declined to disclose “how far along we are on a decision to seek legislation” but leaned forward on the issue.

“I’ve never seen the atmosphere here in D.C. to be so conducive to passing some kind of encryption legislation or lawful access legislation as it is today,” Demers said during a discussion at the Wilson Center.

His remarks come in the wake of last month’s shooting at a naval base in Pensacola, Florida, that killed three people and led the FBI earlier this month to ask Apple for help opening two iPhones that belonged to the Saudi shooter. This week, U.S. Attorney General William P. Barr raised the issue again, accusing Apple of failing to provide “substantial assistance” and calling on the firm “to help us find a solution” to locked devices.

President Trump on Tuesday also weighed in with a harsh tweet: “We are helping Apple all of the time on TRADE and so many other issues, and yet they refuse to unlock phones used by killers, drug dealers and other violent criminal elements,” he said. “They will have to step up to the plate and help our great Country, NOW!”

Demers referred to a Senate hearing in December in which Senate Judiciary Committee Chairman Lindsey Graham, R-S.C., warned tech firms to find a way to build access into their phones or Congress would act. “My advice to you is to get on with it,” he said, “because this time next year, if we haven’t found a way that you can live with, we will impose our will on you.”

Any legislation would still have to pass a Democratic-controlled House, where a bipartisan alliance of privacy hawks and libertarians could block those efforts. Such legislation has been an uphill climb in Congress for years. A bipartisan draft law was circulated after the FBI in early 2016 was unable to get into the iPhone of a terrorist who carried out a mass shooting in San Bernardino, California, but it faced such criticism that it was never introduced.

In a statement, Apple rejected the assertion that it has not provided substantial help in the Pensacola case. Within six hours of the FBI’s first request on Dec. 6 and in the days after, it provided data, including iCloud backups and account information for multiple accounts, the firm said. One account belonged to the shooter.

The FBI notified Apple only on Jan. 6 – a month after the shooting – that it needed additional help, revealing that the gunman had a second phone. But, as the firm argued in 2016 when the FBI wanted help unlocking the phone in the San Bernardino case, it could not break into the device without hacking it. The FBI eventually paid a private contractor $900,000 to crack the passcode after disabling a security feature.

“There’s a good reason why Congress has failed to legislate up to now,” said Jennifer Daskal, a law professor at American University. “Once you get past the talking points, the range of security, privacy and economic risks become apparent.”

The decision amounted to a significant early victory for Maura Healey, the Democratic attorney general of Massachusetts, who said in a statement Friday that Facebook users – and local watchdogs – “have a right to know” whether the company broke the law and violated people’s privacy.

Facebook, however, signaled the fight may not be over, marking its latest effort to battle back state regulators who have intensified their scrutiny of the tech giant.

“We are disappointed that the Massachusetts Attorney General and the Court didn’t fully consider our arguments on well-established law,” spokesman Andy Stone said in a statement, adding: “We are reviewing our options, including appeal.”

Massachusetts revealed it was probing Facebook over its data-collection practices in September, an investigation that stemmed from the company’s entanglement with Cambridge Analytica. That privacy scandal already has resulted in a record-breaking, $5 billion federal fine for Facebook.

The court dispute centered on Facebook’s admission last year that it had suspended “tens of thousands” of apps for possible privacy violations. Facebook discovered the app issues as a result of an internal audit of its third-party developers, but it declined to share – with the public or with Massachusetts officials – exactly who it had suspended or many details about their potential wrongdoings.

Healey and her aides argued the data was critical, potentially showing that thousands of apps, some with large numbers of users, presented an elevated risk of privacy violations or behaved in a way that “may suggest data misuse,” her office said at the time. Facebook, however, fought to keep the evidence to itself, arguing it should be shielded from investigators.

After months of wrangling, the attorney general’s office took the issue before a Suffolk Superior Court judge, who ruled Friday that Facebook must surrender the information. Facebook now has 90 days to comply with the state’s request.

“We are pleased that the Court ordered Facebook to tell our office which other app developers may have engaged in conduct like Cambridge Analytica,” Healey said in a statement.

Facebook, for its part, has fought aggressively against states that have probed its privacy practices in the months after the Federal Trade Commission settled with the company. In California, for example, Facebook’s refusal to turn over key documents prompted Attorney General Xavier Becerra, a Democrat, to take the company to court in November.

Democrats and Republicans at times appeared stunned as they heard tales of tech giants wielding their massive footprints as a weapon, allegedly copying smaller competitors’ features or tweaking their algorithms in ways that put encroaching companies at a costly disadvantage. The testimony came as part of a wide-ranging antitrust probe into Silicon Valley’s biggest players that House lawmakers aim to wrap up – with recommendations for regulation – in the coming months.

“It’s like soccer,” said Kirsten Daru, the general counsel of Tile, which has accused Apple of acting anti-competitively. “You might be the best team in the league, but you’re playing against a team that owns the field, the ball, the stadium and the entire league, and they can change the rules of the game at any time.”

The pleas for regulatory relief resonated with lawmakers, led by Rep. David Cicilline, D-R.I., the chairman of the House’s top antitrust committee. “It has become clear these firms have tremendous power as gatekeepers to shape and control commerce online,” Cicilline said to open the session.

The hearing at the University of Colorado-Boulder put a public face on the pain caused by some of the largest tech companies in the United States. Cicilline and his Democratic and Republican peers have sought to determine if federal antitrust law is sufficient to hold Silicon Valley accountable – and whether changes to federal law are necessary to address anti-competitive concerns in search, smartphones, e-commerce and social networking.

“I think it’s clear there’s abuse in the marketplace and a need for action,” said Rep. Ken Buck, a Republican from Colorado.

The House investigation comes as the U.S. government’s two competition agencies, the Federal Trade Commission and Justice Department, proceed with their own probes into Apple, Amazon, Facebook and Google for potential antitrust violations. Nearly every state attorney general, meanwhile, has trained their sights on Facebook and Google, announcing wide-ranging inquiries of their own earlier this year.

A key leader in those states’ efforts – Colorado Attorney General Phil Weiser – sketched out a broad, ambitious agenda for antitrust enforcement in a private meeting with U.S. lawmakers Friday morning, where he called on them to invest more resources in oversight.

“The idea we’re not going to regulate tech companies is so 1990s,” Weiser said in an interview before he spoke.

At the hearing, Tile took issue with Apple for changes to its most recent iOS software for iPhones and iPads. Tile said Apple’s tools to help smartphone owners find their missing items largely mimics its own offering. Adding to its advantages, Apple imposes tougher restrictions on how Tile and others collect much-needed location data, said Daru, the company’s general counsel.

“Tile welcomes competition,” she said, “but it has to be fair competition.”

Apple says its policies seek only to protect privacy, but lawmakers at times did not appear convinced, pointing to other instances in which the iPhone giant – seeing a successful product in Apple’s app ecosystem – launches competing services of their own.

“Once Apple has essentially decided to do the same, it renders all of those apps superfluous and unnecessary,” said Rep. Joe Neguse, D-Colo.

Patrick Spence, the leader of Sonos, blasted Google and the rest of the industry for “using their power in one market to conquer or destroy nascent markets.” The high-end speaker company alleges in a lawsuit that Google unlawfully copied its technology, a charge Google denies.

PopSockets, a Boulder-based company that makes circular grips for smartphones, took issue with Amazon. David Barnett, the company’s founder, fretted about restrictions Amazon places on sellers. He said PopSockets at one point tried to quit selling through the e-commerce giant, but severing those ties ultimately cost his company $10 million. Amazon has disputed Barnett’s claims.

And David Heinemeier Hansson, the co-founder of Basecamp, which makes Web-based product management tools, said the digital ecosystem as a whole had been “colonized by a handful of big tech companies.” He likened the current behaviors of Apple, Amazon, Facebook and Google to some of the same practices that decades ago led the government to try to penalize Microsoft for antitrust abuses.

“Help us, Congress,” Hansson said, “you’re our only hope.”

The EU’s executive arm is set to propose the new rules apply to “high-risk sectors,” such as healthcare and transport, and suggest the bloc updates safety and liability laws, according to a draft of a so-called “white paper” on artificial intelligence obtained by Bloomberg. The European Commission is due to unveil the paper in mid-February and the final version is likely to change.

The paper is part of the EU’s broader effort to catch up to the U.S. and China on advancements in AI, but in a way that promotes European values such as user privacy. While some critics have long argued that stringent data protection laws like the EU’s could hinder innovation around AI, EU officials say harmonizing rules across the region will boost development.

European Commission President Ursula von der Leyen has pledged her team would present a new legislative approach on artificial intelligence within the first 100 days of her mandate, which started Dec. 1, handing the task to the EU’s digital chief, Margrethe Vestager, to coordinate.

A spokesman for the Brussels-based Commission declined to comment on leaks but added: “To maximize the benefits and address the challenges of Artificial Intelligence, Europe has to act as one and will define its own way, a human way. Trust and security of EU citizens will therefore be at the center of the EU’s strategy.”

The EU is also considering new obligations for public authorities around the deployment of facial recognition technology and more detailed rules on the use of such systems in public spaces. However, the provision on facial recognition isn’t among the three policy options officials recommend that the commission pursue.

The provision suggests prohibiting use of facial recognition by public and private actors in public spaces for several years to allow time to assess the risks of such technology.

“Such a ban would be a far-reaching measure that might hamper the development and uptake of this technology,” the commission says in the document, adding that it’s therefore preferable to focus on implementing relevant provisions in the EU’s existing data protection laws.

As part of the recommended policy measures, the EU also wants to urge its member states to appoint authorities to monitor the enforcement of any future rules governing the use of AI, according to the document.

In the draft, the EU defines high-risk applications as “applications of artificial intelligence which can produce legal effects for the individual or the legal entity or pose risk of injury, death or significant material damage for the individual or the legal entity.”

Artificial intelligence is already subject to a variety of European regulations, including rules on fundamental rights around privacy, non-discrimination, as well as product safety and liability laws, but the rules may not fully cover all specific risks posed by new technologies, the Commission says in the document. For instance, product safety laws currently wouldn’t apply to services based on AI.

The company, founded by entrepreneur Richard Branson, plans to welcome its initial customers later this year in New Mexico for its first commercial flight.

Investors are also betting that Virgin Galactic will be one of the first companies to offer point-to-point hypersonic travel, one day potentially reducing intercontinental flights to less than three hours. In a December report, Morgan Stanley analysts valued that market at $800 billion by 2040, dwarfing the space tourism business.

Virgin announced last week that its second commercial ship had reached a “weight on wheels” assembly milestone considerably faster than it took to get to the same stage with its first spaceship. Work on a third ship has begun, the company said.

Virgin Galactic aims to have five spacecraft in service by the end of 2023 operating from its base at Spaceport America in southern New Mexico.

On Wednesday, the company named Enrico Palermo as its chief operating officer. Palermo was president of its manufacturing unit, The Spaceship Co.

Virgin Galactic rose 1.4% to $15.05 at 1:32 p.m. Thursday in trading on the Nasdaq.

The company debuted on the public markets on Oct. 28 and rose to more than $12 before dipping to a low of $7.22 in late November. The stock began rising again in mid-December, finishing the year at $11.55. Morgan Stanley assigned a $22 target price for the shares on Dec. 19.

The jump last month came after the management team made the rounds of analysts to tell its story, Alex King, founder of Cestrian Capital Research, wrote in a Jan. 13 note. King owns Virgin Galactic shares personally.

In November, Virgin reported losing about $128 million for the period through Sept. 30. He labeled the stock “speculative” given the company’s early stage. Virgin Galactic has raised more than $1 billion since it was founded in 2004, initially from Branson, with an Abu Dhabi investment company taking a stake in 2010.

The software giant, which announced the initiative at a news conference at its Redmond headquarters, also plans to remove all the carbon emissions it has generated since its founding in 1975 by 2050.

“The scientific consensus is clear,” Microsoft chief executive Satya Nadella said at an event Thursday morning at Microsoft’s headquarters. “The world today is confronted with an urgent carbon crisis. Each of us is going to need to take action, and that includes business.”

Microsoft’s initiative goes a few steps beyond what crosstown rival Amazon announced in September. The e-commerce giant, which competes with Microsoft in the booming business of cloud computing, said at the time it would implement strategies to be at net zero emissions – removing as much carbon as it produces – throughout its business by 2040.

Working to remove carbon emissions to prevent further global warming is “far and away the biggest challenge, I think, that humanity has ever been presented with,” Microsoft President Brad Smith said in a meeting with journalists Monday in New York.

Microsoft’s Smith credited Amazon, but noted that Microsoft’s ambitions for addressing climate change are bolder. “They said they’ll be carbon-neutral by 2040. We’ve said we’ll be carbon-negative by 2030,” Smith said. “They’ve said they’ll have 100 percent renewable energy for their data centers by 2030. We’ve said we’ll do that by 2025.”

(Amazon chief executive Jeff Bezos owns The Washington Post.)

One point on which Microsoft and Amazon are aligned is that they both will continue to provide artificial intelligence and machine learning technology to energy companies to help them better pinpoint oil deposits for extraction.

Activist employees at Amazon have protested their employer’s work with energy companies, so much so that a lawyer in the company’s human resources group raised the possibility of termination if they didn’t adhere to its external communications policy. Last month, an anonymous author who claimed to work for Microsoft lamented the company’s work with energy firms in an article in Logic Magazine. The author recounted a work trip to Kazakhstan to help the state-owned energy company work with Chevron to extract oil from a field near the city of Atyrau.

“Microsoft executives aren’t going to give up on the billions of dollars to be made from Big Oil, especially if it helps them win more of the coveted cloud market,” the author wrote.

Microsoft declined to comment on the article, but Smith said in an interview Thursday that the company intends to continue working with oil and gas companies, in part to help them move to cleaner forms of energy.

“We are a company that wants to solve problems by working with people and by helping them make the transitions the world needs to make, rather than by cutting them off and denying them tech writ large,” Smith said.

Microsoft previously pledged to continue working with the U.S. military after some employees protested some contracts with Pentagon, saying it wanted to be part of the discussion about using technology responsibly. Microsoft applied the same logic to its decision to work with oil and gas companies.

“We are consistent philosophically, especially when it’s a vertical sector that can be part of a solution,” Smith said.

In working toward being carbon-negative by 2030, Microsoft is counting emissions it produces, such as exhaust from its fleet of cars and trucks, as well as emissions it indirectly causes, such as those that come from electricity it uses, the production of the materials in its buildings, and the power customers consume when using its products, such as the Xbox game console. But Microsoft won’t count the emissions from energy produced by oil and gas companies that its technology helped find.

To achieve its goal, Microsoft plans to have its entire fleet of vehicles run on electric power by 2030. It will adopt so-called negative emission technologies including soil carbon sequestration and bioenergy with carbon capture and storage to remove emissions it’s created. The company also committed to investing $1 billion over the next four years in new technologies to help address the climate change crisis.

And while Microsoft said it will continue to disclose its carbon footprint, Smith said the company hasn’t committed to an independent audit of the progress against its goals. The company will push for better standards to measure and report carbon emissions and removal.

“Whether it actually needs to be audited, we’ll see,” Smith said.

He declined to disclose the cost of the new carbon-reduction initiative.

Other tech giants have dialed up efforts to reduce their emissions, including Apple, which last spring said that 21 manufacturers in its supply chain vowed to obtain all their electricity from renewable sources, a number the company now says is up to 44.

The public disclosure represents a major shift in the NSA’s approach, choosing to put computer security ahead of building up its arsenal of hacking tools that allow the agency to spy on adversaries’ networks.

“This is . . . a change in approach . . . by NSA of working to share, working to lean forward, and then working to really share the data as part of building trust,” said Anne Neuberger, director of the NSA’s Cybersecurity Directorate, which was launched in October.

Cyber security professionals hailed the move.

“Big kudos to NSA for voluntarily disclosing to Microsoft,” said computer security expert Dmitri Alperovitch in a tweet Tuesday morning. “This is the type of [vulnerability] I am sure the [NSA hackers] would have loved to use for years to come.”

The bug – essentially a mistake in the computer code – affects the Windows 10 operating system, the most widely used in government and business today.

Microsoft issued a patch for the flaw on Tuesday. The company’s plan to issue a fix for the vulnerability was first reported Monday in the KrebsOnSecurity blog.

“A security update was released on January 14, 2020 and customers who have already applied the update, or have automatic updates enabled, are already protected. As always we encourage customers to install all security updates as soon as possible,” said Jeff Jones, senior director at Microsoft, in a statement.

The discovery has been likened to a slightly less severe version of the Microsoft flaw that the NSA once weaponized by creating a hacking tool dubbed EternalBlue, which one former agency hacker said was like “fishing with dynamite.”

The NSA used EternalBlue for more than five years, but when it learned that the tool had been obtained by others, it alerted Microsoft, which issued a patch in early 2017. About a month later, Shadow Brokers, a suspected Russian hacking group, released the NSA tool online.

Malicious hackers turned it to their own purposes, launching massive ransomware campaigns such as the one dubbed WannaCry, which created global havoc and costly damage to businesses and other organizations.

EternalBlue worked on all Windows systems, not just one, which made it so potent. The flaw the NSA uncovered would be useful to hackers seeking to break into some computers running Windows 10.

Companies like Microsoft and Adobe use digital signatures to stamp software as authentic. This helps to prevent malware infections that might try to disguise themselves as legitimate. The NSA discovered an error in the Microsoft code that verifies those signatures, potentially enabling a hacker to forge the signature and install spyware or ransomware on a computer.

“Code-signing is one of the most effective tools we have to keep malicious software off of computers,” said Matthew Green, a cryptographer and computer science professor at Johns Hopkins University.

If the flaw is patched quickly, it’s not that dangerous, he added. “If a lot of people don’t patch, it could be a disaster.”

Microsoft has reported that it has seen no active exploitation of the flaw.

The bug disclosure is the first major announcement to come from the new directorate, which reflects NSA Director Gen. Paul Nakasone’s desire to enhance the defensive mission of an agency known for its prowess at hacking foreign networks for intelligence.

Over the next two years the Alphabet Inc. unit intends to stop supporting third-party cookies in its Chrome browser, Google said in a blog post Tuesday. Cookies are little bits of code that stick in peoples’ browsers and follow them around the web and are a core part of the online advertising landscape. They allow advertisers to target people with ads for websites they previously visited and make it easier to determine how effective certain ads were in getting internet surfers to buy something.

Apple Inc.’s Safari and Mozilla Corp.’s Firefox browsers already block third-party cookies, but Google has argued in the past those approaches are too heavy-handed and risk cutting into vital revenue for internet publishers. In the blog post, Chrome Engineering Director Justin Schuh said blocking third-party cookies could have “unintended consequences that can negatively impact both users and the web ecosystem.” The company is seeking input from advertisers, publishers and Chrome users as it works to find ways to help support advertising online while still preserving privacy.

“Users are demanding greater privacy — including transparency, choice and control over how their data is used — and it’s clear the web ecosystem needs to evolve to meet these increasing demands,” Schuh said.

Google has talked about this approach before. While Apple and Mozilla don’t derive much money from advertising, the vast majority of Google’s revenue comes from digital ads. It’s in the company’s interest to keep advertisers spending more money on its websites and ad products. Google’s empire was built on its ability to provide targeted advertising.

Google is navigating a thicket of threats to its business though, including a rising demand for greater privacy and government investigations into whether its business practices in the ad tech world are anti-competitive. If it shuts advertisers out from its system too much, they could increase their complaints that it’s being unfair. But if it ignores privacy advocates, some Chrome users could decamp for other browsers.