Світ

The maker of ChatGPT is trying to curb its reputation as a freewheeling cheating machine with a new tool that can help teachers detect if a student or artificial intelligence wrote that homework.

The new AI Text Classifier launched Tuesday by OpenAI follows a weeks-long discussion at schools and colleges over fears that ChatGPT’s ability to write just about anything on command could fuel academic dishonesty and hinder learning.

OpenAI cautions that its new tool – like others already available – is not foolproof. The method for detecting AI-written text “is imperfect and it will be wrong sometimes,” said Jan Leike, head of OpenAI’s alignment team tasked to make its systems safer.

“Because of that, it shouldn’t be solely relied upon when making decisions,” Leike said.

Teenagers and college students were among the millions of people who began experimenting with ChatGPT after it launched November 30 as a free application on OpenAI’s website. And while many found ways to use it creatively and harmlessly, the ease with which it could answer take-home test questions and assist with other assignments sparked a panic among some educators.

By the time schools opened for the new year, New York City, Los Angeles and other big public school districts began to block its use in classrooms and on school devices.

The Seattle Public Schools district initially blocked ChatGPT on all school devices in December but then opened access to educators who want to use it as a teaching tool, said Tim Robinson, the district spokesman.

“We can’t afford to ignore it,” Robinson said.

The district is also discussing possibly expanding the use of ChatGPT into classrooms to let teachers use it to train students to be better critical thinkers and to let students use the application as a “personal tutor” or to help generate new ideas when working on an assignment, Robinson said.

School districts around the country say they are seeing the conversation around ChatGPT evolve quickly.

“The initial reaction was ‘OMG, how are we going to stem the tide of all the cheating that will happen with ChatGPT,'” said Devin Page, a technology specialist with the Calvert County Public School District in Maryland. Now there is a growing realization that “this is the future” and blocking it is not the solution, he said.

“I think we would be naïve if we were not aware of the dangers this tool poses, but we also would fail to serve our students if we ban them and us from using it for all its potential power,” said Page, who thinks districts like his own will eventually unblock ChatGPT, especially once the company’s detection service is in place.

OpenAI emphasized the limitations of its detection tool in a blog post Tuesday, but said that in addition to deterring plagiarism, it could help to detect automated disinformation campaigns and other misuse of AI to mimic humans.

The longer a passage of text, the better the tool is at detecting if an AI or human wrote something. Type in any text — a college admissions essay, or a literary analysis of Ralph Ellison’s “Invisible Man” — and the tool will label it as either “very unlikely, unlikely, unclear if it is, possibly, or likely” AI-generated.

But much like ChatGPT itself, which was trained on a huge trove of digitized books, newspapers and online writings but often confidently spits out falsehoods or nonsense, it’s not easy to interpret how it came up with a result.

“We don’t fundamentally know what kind of pattern it pays attention to, or how it works internally,” Leike said. “There’s really not much we could say at this point about how the classifier actually works.”

“Like many other technologies, it may be that one district decides that it’s inappropriate for use in their classrooms,” said OpenAI policy researcher Lama Ahmad. “We don’t really push them one way or another.”
…

China says it is “deeply concerned” over reports that the United States is moving to further restrict sales of American technology to Huawei, a tech company that U.S. officials have long singled out as a threat to national security for its alleged support of Beijing’s espionage efforts.

As first reported by the Financial Times, the U.S. Department of Commerce has informed American firms that it will no longer issue licenses for technology exports to Huawei, thereby isolating the Shenzen-based company from supplies it needs to make its products.

The White House and Commerce Department have not responded to VOA’s request for confirmation of the reports. But observers say the move may be the latest tactic in the Biden administration’s geoeconomics strategy as it comes under increasing Republican pressure to outcompete China. 

The crackdown on Chinese companies began under the Trump administration, which in 2019 added Huawei to an export blacklist but made exceptions for some American firms, including Qualcomm and Intel, to provide non-5G technology licenses.

Since taking office in 2021, President Joe Biden has taken an even more aggressive stance than his predecessor, Donald Trump. Now the Biden administration appears to be heading toward a total ban on all tech exports to Huawei, said Sam Howell, who researches quantum information science at the Center for a New American Security’s Technology and National Security program.

“These new restrictions from what we understand so far would include items below the 5G level,” she told VOA. “So 4G items, Wi-Fi 6 and [Wi-Fi] 7, artificial intelligence, high performance computing and cloud capabilities as well.”

Should the Commerce Department follow through with the ban, there will likely be pushback from U.S. companies whose revenues will be directly affected, Howell said. Currently Intel and Qualcomm still sell chips used in laptops and phones manufactured by Huawei.

Huawei and Beijing have denied that they are a threat to other countries’ national security. Foreign ministry spokesperson Mao Ning accused Washington of “overstretching the concept of national security and abusing state power” to suppress Chinese competitors.

“Such practices are contrary to the principles of market economy” and are “blatant technological hegemony,” Mao said. 

Outcompeting Chinese tech

The latest U.S. move on Huawei is part of a U.S. effort to outcompete China in the cutting-edge technology sector.

In October, Biden imposed sweeping restrictions on providing advanced semiconductors and chipmaking equipment to Chinese companies, seeking to maintain dominance particularly on the most advanced chips. His administration is rallying allies behind the effort, including the Netherlands, Japan, South Korea and Taiwan – home to leading companies that play key roles in the industry’s supply chain.

U.S. officials say export restrictions on chips are necessary because China can use semiconductors to advance their military systems, including weapons of mass destruction, and commit human rights abuses. 

The October restrictions follow the CHIPS and Science Act of 2022, which Biden signed into law in August and that restricts companies receiving U.S. subsidies from investing in and expanding cutting-edge chipmaking facilities in China. It also provides $52 billion to strengthen the domestic semiconductor industry.

Beijing has invested heavily in its own semiconductor sector, with plans to invest $1.4 trillion in advanced technologies in a bid to achieve 70% self-sufficiency in semiconductors by 2025. 

TikTok a target

TikTok, a social media application owned by the Chinese company ByteDance that has built a massive following especially among American youth, is also under U.S. lawmakers’ scrutiny due to suspicion that it could be used as a tool of Chinese foreign espionage or influence.

CEO Shou Zi Chew is scheduled to appear before the House Energy and Commerce Committee on March 23 to testify about TikTok’s “consumer privacy and data security practices, the platforms’ impact on kids, and their relationship with the Chinese Communist Party.”

Lawmakers are divided on whether to ban or allow the popular app, which has been downloaded onto about 100 million U.S. smartphones, or force its sale to an American buyer.

Earlier in January, Congress set up the House Select Committee on China, tasked with dealing with legislation to combat the dangers of a rising China.
…

For New York teacher Michael Flanagan, the pandemic was a crash course in new technology — rushing out laptops to stay-at-home students and shifting hectic school life online.

Students are long back at school, but the technology has lived on, and with it has come a new generation of apps that monitor the pupils online, sometimes round the clock and even on down days shared with family and friends at home.

The programs scan students’ online activity, social media posts and more — aiming to keep them focused, detect mental health problems and flag up any potential for violence.

“You can’t unring the bell,” said Flanagan, who teaches social studies and economics. “Everybody has a device.”

The new trend for tracking, however, has raised fears that some of the apps may target minority pupils, while others have outed LGBT+ students without their consent, and many are used to instill discipline as much as deliver care.

So Flanagan has parted ways with many of his colleagues and won’t use such apps to monitor his students online.

He recalled seeing a demo of one such program, GoGuardian, in which a teacher showed — in real time — what one student was doing on his computer. The child was at home, on a day off.

Such scrutiny raised a big red flag for Flanagan.

“I have a school-issued device, and I know that there’s no expectation of privacy. But I’m a grown man — these kids don’t know that,” he said.

A New York City Department of Education spokesperson said that the use of GoGuardian Teacher “is only for teachers to see what’s on the student’s screen in the moment, provide refocusing prompts, and limit access to inappropriate content.”

Valued at more than $1 billion, GoGuardian — one of a handful of high-profile apps in the market — is now monitoring more than 22 million students, including in the New York City, Chicago and Los Angeles public systems.

Globally, the education technology sector is expected to grow by $133 billion from 2021 to 2026, market researcher Technavio said last year.

Parents expect schools to keep children safe in classrooms or on field trips, and schools also “have a responsibility to keep students safe in digital spaces and on school-issued devices,” GoGuardian said in a statement.

The company says it “provides educators with the ability to protect students from harmful or explicit content”.

Nowadays, online monitoring “is just part of the school environment,” said Jamie Gorosh, policy counsel with the Future of Privacy Forum, a watchdog group.

And even as schools move beyond the pandemic, “it doesn’t look like we’re going back,” she said.

Guns and depression

A key priority for monitoring is to keep students engaged in their academic work, but it also taps into fast-rising concerns over school violence and children’s mental health, which medical groups in 2021 termed a national emergency.

According to federal data released this month, 82% of schools now train staff on how to spot mental health problems, up from 60% in 2018; 65% have confidential threat-reporting systems, up 15% in the same period.

In a survey last year by the nonprofit Center for Democracy and Technology (CDT), 89% of teachers reported their schools were monitoring student online activity.

Yet it is not clear that the software creates safer schools.

Gorosh cited May’s shooting in Uvalde, Texas, that left 21 dead in a school that had invested heavily in monitoring tech.

Some worry the tracking apps could actively cause harm.

The CDT report, for instance, found that while administrators overwhelmingly say the purpose of monitoring software is student safety, “it’s being used far more commonly for disciplinary purposes … and we’re seeing a discrepancy falling along racial lines,” said Elizabeth Laird, director of CDT’s Equity in Civic Technology program.

The programs’ use of artificial intelligence to scan for keywords has also outed LGBT+ students without their consent, she said, noting that 29% of students who identify as LGBT+ said they or someone they knew had experienced this.

And more than a third of teachers said their schools send alerts automatically to law enforcement outside school hours.

“The stated purpose is to keep students safe, and here we have set up a system that is routinizing law enforcement access to this information and finding reasons for them to go into students’ homes,” Laird said.

‘Preyed upon’

A report by federal lawmakers last year into four companies making student monitoring software found that none had made efforts to see if the programs disproportionately targeted marginalized students.

“Students should not be surveilled on the same platforms they use for their schooling,” Senator Ed Markey of Massachusetts, one of the report’s co-authors, told the Thomson Reuters Foundation in a statement.

“As school districts work to incorporate technology in the classroom, we must ensure children and teenagers are not preyed upon by a web of targeted advertising or intrusive monitoring of any kind.”

The Department of Education has committed to releasing guidelines around the use of AI early this year.

A spokesperson said the agency was “committed to protecting the civil rights of all students.”

Aside from the ethical questions around spying on children, many parents are frustrated by the lack of transparency.

“We need more clarity on whether data is being collected, especially sensitive data. You should have at least notification, and probably consent,” said Cassie Creswell, head of Illinois Families for Public Schools, an advocacy group.

Creswell, who has a daughter in a Chicago public school, said several parents have been sent alerts about their children’s online searches, despite not having been asked or told about the monitoring in the first place.

Another child had faced repeated warnings not to play a particular game — even though the student was playing it at home on the family computer, she said.

Creswell and others acknowledge that the issues monitoring aims to address — bullying, depression, violence — are real and need tackling, but question whether technology is the answer.

“If we’re talking about self-harm monitoring, is this the best way to approach the issue?” said Gorosh.

Pointing to evidence suggesting AI is imperfect in capturing the warning signs, she said increased funding for school counselors could be more narrowly tailored to the problem.

“There are huge concerns,” she said. “But maybe technology isn’t the first step to answer some of those issues.”
…

The United States and European Union announced Friday an agreement to speed up and enhance the use of artificial intelligence to improve agriculture, health care, emergency response, climate forecasting and the electric grid. 

A senior U.S. administration official, discussing the initiative shortly before the official announcement, called it the first sweeping AI agreement between the United States and Europe. Previously, agreements on the issue had been limited to specific areas such as enhancing privacy, the official said.  

AI modeling, which refers to machine-learning algorithms that use data to make logical decisions, could be used to improve the speed and efficiency of government operations and services.  

“The magic here is in building joint models [while] leaving data where it is,” the senior administration official said. “The U.S. data stays in the U.S. and European data stays there, but we can build a model that talks to the European and the U.S. data, because the more data and the more diverse data, the better the model.” 

The initiative will give governments greater access to more detailed and data-rich AI models, leading to more efficient emergency responses and electric grid management, and other benefits, the administration official said. 

Pointing to the electric grid, the official said the United States collects data on how electricity is being used, where it is generated, and how to balance the grid’s load so that weather changes do not knock it offline. 

Many European countries have similar data points they gather relating to their own grids, the official said. Under the new partnership, all that data would be harnessed into a common AI model that would produce better results for emergency managers, grid operators and others relying on AI to improve systems.  

The partnership is currently between the White House and the European Commission, the executive arm of the 27-member European Union. The senior administration official said other countries would be invited to join in the coming months.  
…

An international ransomware network that extorted more than $100 million from hospitals and other organizations around the world has been brought down following a monthslong infiltration by the FBI, the Justice Department said Thursday.

The Hive ransomware group, known to operate since June 2021, targeted more than 1,500 victims, including hospitals, school districts and financial firms in more than 80 countries, DOJ and FBI officials said at a press conference. The network’s most recent victim in Florida was targeted about two weeks ago.

FBI agents, who penetrated the group’s computer networks last summer and thwarted multiple attacks, seized its two Los Angeles-based servers  Wednesday night, while taking control of darknet sites used by its affiliates, officials said.

German and Dutch police took part in the international law enforcement action.

Attorney General Merrick Garland and other top law enforcement officials announced the operation.

“Cybercrime is a constantly evolving threat,” Garland said. “But as I have said before, the Justice Department will spare no resource to identify and bring to justice anyone, anywhere, who targets the United States with a ransomware attack.”

In a ransomware attack, hackers encrypt the data on a victim’s network and then demand payments in exchange for providing a decryption key.

Hive used a “ransomware-as-a-service” model in which highly skilled developers build the malware and then recruit less-sophisticated affiliates to deploy them against victims.

Garland said Hive affiliates targeted “critical infrastructure and some of our nation’s most important industries.”

In August 2021, at the height of the COVID-19 pandemic, Hive affiliates attacked a Midwest hospital’s network, preventing the medical facility from accepting new patients, Garland said.

The hospital was able to recover its data only after paying a ransom, the attorney general said.

While no arrests have been made in connection with the operation, FBI Director Christopher Wray warned that “anybody involved with Hive should be concerned, because this investigation is very much ongoing.”

“We’re engaged in what we call ‘joint sequenced operations’ … and that includes going after their infrastructure, going after their crypto and going after the people who work with them,” Wray said.

FBI agents infiltrated Hive from July 2022 until its seizure, covertly capturing its decryption keys and sharing them with victims, saving the targets $130 million in ransom payments, officials said.

“Simply put, using lawful means, we hacked the hackers,” Deputy Attorney General Lisa Monaco said.

In all, the FBI provided more than 300 victims with decryption keys, Garland said, among them a Texas school district, a Louisiana hospital, and a food services company that had been asked to make millions of dollars in ransom payments. The FBI also distributed more than 1,000 additional decryption keys to previous Hive victims.

The takedown represents a win for the Biden administration’s efforts to crack down on a recent surge in ransomware attacks that cost businesses and governments around the world billions of dollars a year.

U.S. banks and financial institutions processed nearly $1.2 billion in suspected ransomware payments in 2021, more than double the amount in 2020, the Treasury Department’s Financial Crimes Enforcement Network (FinCen) reported in November.

Roughly 75% of the ransomware attacks reported in 2021 had a nexus with Russia, its proxies or persons acting on its behalf, according to FinCen, which also says the top five highest-grossing ransomware tools used in 2021 were all connected to Russian cyberactors.

Officials would not say whether Hive had any known links to Russia.

John Bennett, a former senior FBI official who is now managing director of the Cyber Risk Business Unit at Kroll, a cybersecurity services company, noted that the seizure notice on Hive’s website, written in both English and a Slavic language, suggests it is aimed at an Eastern European audience.

“The fact that it is basically being broadcast in a [Slavic] language, I think, is telling that that’s the target audience that they’re letting know that they got this,” Bennett said in an interview.

The gang’s takedown, Bennett said, is a sign of what is coming.

“I think this is telling that law enforcement is catching up very quickly to the capabilities of getting inside of these groups,” Bennett said.
…