By Reuters | Updated: 1 February 2023

OpenAI, the creator of the popular chatbot ChatGPT, has released a software tool to identify text generated by artificial intelligence, the company said in a blog post on Wednesday.

ChatGPT is a free program that generates text in response to a prompt, including articles, essays, jokes, and even poetry, which has gained wide popularity since its debut in November, while raising concerns about copyright and plagiarism.

The AI classifier, a language model trained on the dataset of pairs of human-written and AI-written text on the same topic, aims to distinguish text that is written by AI. It uses a variety of providers to address issues such as automated misinformation campaigns and academic dishonesty, the company said.

In its public beta mode, OpenAI acknowledges the detection tool is very unreliable on texts under 1,000 characters, and AI-written text can be edited to trick the classifier.

“We’re making this classifier publicly available to get feedback on whether imperfect tools like this one are useful,” OpenAI said.

“We recognize that identifying AI-written text has been an important point of discussion among educators, and equally important is recognizing the limits and impacts of AI-generated text classifiers in the classroom.”

Since ChatGPT debuted in November and gained wide popularity among millions of users, some of the largest US school districts, including New York City, have banned the AI chatbot over concerns that students will use the text generator to cheat or plagiarize.

Others have created third-party detection tools including GPTZeroX to help educators detect AI-generated text.

OpenAI said it is engaging with educators to discuss ChatGPT’s capabilities and limitations and will continue to work on the detection of AI-generated text.

© Thomson Reuters 2023

By Reuters | Updated: 31 January 2023

Facebook on Monday asked a London tribunal to block a collective lawsuit valued at up to GBP 3 billion (roughly Rs. 30,300 crore) over allegations the social media giant abused its dominant position to monetise users’ personal data.

Meta, the parent company of the Facebook group, is facing a mass action brought on behalf of around 45 million Facebook users in Britain.

Legal academic Liza Lovdahl Gormsen, who is bringing the case, said Facebook users were not properly compensated for the value of personal data that they had to provide to use the platform.

Her lawyers said users should get compensation for the economic value they would have received if Facebook was not in a dominant position in the market for social networks.

But Meta said the lawsuit was “entirely without merit” and should not be allowed to proceed. Its lawyers said the claimed losses ignore the “economic value” Facebook provides.

Lovdahl Gormsen’s lawyers on Monday asked the Competition Appeal Tribunal to certify the case under the UK’s collective proceedings regime – which is roughly equivalent to the class action regime in the United States.

A decision to certify collective proceedings will depend on whether the tribunal decides that the individual cases can appropriately be dealt with together, rather than on their merits.

Ronit Kreisberger, representing Lovdahl Gormsen, told the tribunal that “Meta’s data practices violate the prohibition on abusive conduct by dominant firms”.

“There is unquestionably a case for Meta to answer at trial,” Kreisberger argued.

But lawyers representing Meta said the lawsuit wrongly assumes that any “excess profits” it might make equates to a financial loss suffered by individual Facebook users.

This approach “takes no account whatsoever of the significant economic value of the service provided by Facebook”, Marie Demetriou said in court documents.

She said Lovdahl Gormsen’s estimate of potential claimants’ total losses – GBP 3 billion, including interest – is “at the very least wildly inflated”.

© Thomson Reuters 2023

By Reuters | Updated: 31 January 2023

Twitter is working to introduce payments on the social media platform and has begun applying for regulatory licenses, the Financial Times reported on Monday, citing people familiar with the matter.

New boss Elon Musk is pushing Twitter to create new streams of revenue as it faces a drop in advertising income, following his $44 billion (roughly Rs. 3.6 lakh crore) takeover of the company in October.

The development of the payments feature is being led by Esther Crawford, a director of product management at Twitter, according to the report, which added that the executive was emerging to be a key lieutenant to Musk.

Twitter did not immediately respond to a Reuters request for comment.

Musk had previously said that the Twitter acquisition would be part of a master plan to create “the everything app”, a service that would offer social networking, peer-to-peer payments and e-commerce shopping.

Prior to Musk’s takeover, Twitter in early 2021 was exploring allowing its users to receive tips, or digital payments, from their followers.

Meanwhile, Twitter announced last week that users will be able to appeal account suspensions and be evaluated under the social media platform’s new criteria for reinstatement, starting February 1.

Under the new criteria, which follow billionaire Elon Musk’s purchase of the company in October, Twitter accounts will only be suspended for severe or ongoing and repeat violations of the platform’s policies.

Severe policy violations include engaging in illegal content or activity, inciting or threatening violence or harm, and engaging in targeted harassment of other users, among others.

Twitter said that going forward, it will take less severe action, in comparison to account suspension, such as limiting the reach of tweets that violate its policies or asking users to remove tweets before continuing to use the account.

© Thomson Reuters 2023

By Reuters | Updated: 31 January 2023

A US judge on Monday said the names of two people who helped guarantee bail for indicted FTX cryptocurrency exchange founder Sam Bankman-Fried should be made public, but put his ruling on hold pending an expected appeal.

US District Judge Lewis Kaplan in Manhattan ruled in favour of several media outlets including Reuters that sought the names.

The judge said that while the public had only a “weak” right to know who Bankman-Fried’s guarantors were, it outweighed Bankman-Fried’s arguments for confidentiality, including that the guarantors’ safety could be imperilled.

Kaplan also said the names will remain under seal until at least February 7, because “the question presented here is novel and an appeal is likely.” A spokesman for Mark Cohen and Christian Everdell, who represent Bankman-Fried, declined to comment. Bankman-Fried, 30, has been confined at his parents’ home in California, after pleading not guilty to fraud for allegedly looting billions of FTX customer dollars.

His parents, both professors at Stanford Law School, had co-signed a $250 million (roughly Rs. 2,041 crore) bond for their son, with two other guarantors required to sign $500,000 (roughly Rs. 4 crore) and $200,000 (roughly Rs. 1.6 crore) bonds.

Bankman-Fried’s lawyers said the parents had been harassed and received physical threats since FTX’s November collapse and bankruptcy, and there was “serious cause for concern” the additional guarantors might suffer similar treatment.

Kaplan disagreed, noting that long before bail was posted, the parents had faced “intense public scrutiny” over their relationship with their son, who was once worth an estimated $26 billion (roughly Rs. 2 lakh crore).

“The amounts of the individual bonds — $500,000 and $200,000 — do not suggest that the non-parental sureties are persons of great wealth or likely to attract the attention of the types and volume of that to which defendant’s parents appear to have been subjected,” Kaplan wrote.

Media outlets distinguished the case from another judge’s decision not to reveal who guaranteed a bond for Jeffrey Epstein’s longtime associate Ghislaine Maxwell.

They said there was less “stigma” from being associated with Bankman-Fried than from being associated with the late sex offender. Maxwell was later convicted.

Other media seeking to identify Bankman-Fried’s guarantors included the Associated Press, Bloomberg, CNBC, CoinDesk, Dow Jones, the Financial Times, Insider, the New York Times and the Washington Post.

© Thomson Reuters 2023

By Reuters | Updated: 31 January 2023 10:03 IST

The Biden administration has stopped approving licenses for US companies to export most items to China’s Huawei, according to three people familiar with the matter.

Huawei has faced US export restrictions around items for 5G and other technologies for several years, but officials in the US Department of Commerce have granted licenses for some American firms to sell certain goods and technologies to the company. Qualcomm, in 2020, received permission to sell 4G smartphone chips to Huawei.

A Commerce Department spokesperson said officials “continually assess our policies and regulations” but do not comment on talks with specific companies. Huawei and Qualcomm declined to comment. Bloomberg and the Financial Times earlier reported the move.

One person familiar with the matter said US officials are creating a new formal policy of denial for shipping items to Huawei that would include items below the 5G level, including 4G items, Wi-Fi 6 and 7, artificial intelligence, and high-performance computing and cloud items.

Another person said the move was expected to reflect the Biden administration’s tightening of policy on Huawei over the past year. Licenses for 4G chips that could not be used for 5G, which might have been approved earlier, were being denied, the person said. Toward the end of the Trump administration and early in the Biden administration, officials had still granted licenses for items specific to 4G applications.

American officials placed Huawei on a trade blacklist in 2019 restricting most US suppliers from shipping goods and technology to the company unless they were granted licenses. Officials continued to tighten the controls to cut off Huawei’s ability to buy or design the semiconductor chips that power most of its products.

But US officials granted licenses that allowed Huawei to receive some products. For example, suppliers to Huawei got licenses worth $61 billion (roughly Rs. 5 lakh crore) to sell to the telecoms equipment giant from April through November 2021.

In December, Huawei said its overall revenue was about $91.53 billion (roughly Rs. 7.5 lakh crore), down only slightly from 2021 when US sanctions caused its sales to fall by nearly a third.

© Thomson Reuters 2023

By Reuters | Updated: 30 January 2023

Chinese Internet giant Baidu is planning to launch an artificial intelligence chatbot tool similar to OpenAI’s ChatGPT in March, a person familiar with the matter told Reuters.

Baidu plans to debut the application by initially embedding it into its main search services, Bloomberg News reported earlier.

ChatGPT’s tech works by learning from vast amounts of data how to answer any prompt by a user in a human-like way, offering the information like a search engine would or prose like an aspiring novelist.

Microsoft has a $1 billion investment in San Francisco-based OpenAI that it has looked at increasing, Reuters has reported. The company has also worked to add OpenAI’s image-generation software to its Bing search engine in a new challenge to Alphabet Inc’s Google.

Last week, the company announced a further multibillion dollar investment in OpenAI, deepening ties with the startup behind the chatbot sensation ChatGPT and setting the stage for more competition with rival Alphabet Inc’s Google.

Microsoft in a blog post announced “the third phase” of its partnership “through a multiyear, multibillion dollar investment” including additional supercomputer development and cloud-computing support for OpenAI.

Both companies will be able to commercialize the AI tech that results, the blog post said.

A Microsoft spokesperson declined to comment on the terms of the latest investment, which some media outlets earlier reported would be $10 billion (roughly Rs. 82,000 crore).

The widely anticipated investment shows how Microsoft is locked in competition with Google, the inventor of key AI research that is planning its own unveil for this spring, a person familiar with the matter previously told Reuters.

Microsoft’s bet came days after it and Alphabet each announced layoffs of 10,000 or more workers. Redmond, Washington-based Microsoft warned of a recession and growing scrutiny of digital spend by customers in its layoff announcement.

© Thomson Reuters 2023