Connect with us

Tech

Study Finds Most People Can No Longer Tell AI-Generated Voices from Real Ones

Published

on

A new study has found that most people can no longer distinguish between human voices and their artificial intelligence (AI)-generated counterparts, raising growing concerns about misinformation, fraud, and the ethical use of voice-cloning technologies.

The research, published in the journal PLoS One by scientists from Queen Mary University of London, revealed that participants were able to correctly identify genuine human voices only slightly more often than they could identify cloned AI voices. Out of 80 voice samples—half human and half AI-generated—participants mistook 58 percent of cloned voices for real, while 62 percent of actual human voices were correctly identified.

“The most important aspect of the research is that AI-generated voices, specifically voice clones, sound as human as recordings of real human voices,” said Dr. Nadine Lavan, lead author of the study and senior lecturer in psychology at Queen Mary University. She added that these realistic voices were created using commercially available tools, meaning anyone can produce convincing replicas without advanced technical skills or large budgets.

AI voice cloning works by analyzing vocal data to capture and reproduce unique characteristics such as tone, pitch, and rhythm. This precise imitation has made the technology increasingly popular among scammers, who use cloned voices to impersonate loved ones or public figures. According to research by the University of Portsmouth, nearly two-thirds of people over 75 have received attempted phone scams, with about 60 percent of those attempts made through voice calls.

The spread of AI-generated “deepfake” audio has also been used to mimic politicians, journalists, and celebrities, raising fears about its potential to manipulate public opinion and spread false information.

See also  European Commission Closes Better Regulation Consultation, Public Calls for Strong Impact Assessments

Dr. Lavan urged developers to adopt stronger ethical safeguards and work closely with policymakers. “Companies creating the technology should consult ethicists and lawmakers to address issues around voice ownership, consent, and the legal implications of cloning,” she said.

Despite its risks, researchers say the technology also has significant potential for positive impact. AI-generated voices can help restore speech to people who have lost their ability to speak or allow users to design custom voices that reflect their identity.

“This technology could transform accessibility in education, media, and communication,” Lavan noted. She highlighted examples such as AI-assisted audio learning, which has been shown to improve reading engagement among students with neurodiverse conditions like ADHD.

Lavan and her team plan to continue studying how people interact with AI-generated voices, exploring whether knowing a voice is artificial affects trust, engagement, or emotional response.

“As AI voices become part of our daily lives, understanding how we relate to them will be crucial,” she said.

Tech

Anthropic Revenue Jumps to $17 Billion as AI Spending Drives Up Costs Ahead of Nasdaq IPO

Published

on

Anthropic reported a 47 per cent increase in third-quarter revenue to $17 billion (€15.1 billion), but returned to an operating loss as spending on computing capacity and long-term infrastructure commitments mounted ahead of its planned Nasdaq listing.

The US company behind the Claude artificial intelligence models recorded revenue 12 times higher than a year earlier, according to an updated prospectus filed with US regulators on Monday. However, it reported an operating loss of $567 million (€506 million), compared with a profit of $974 million (€870 million) in the previous quarter.

Anthropic attributed part of the reversal to $5 billion (€4.46 billion) spent on short-term computing capacity after demand exceeded the company’s existing long-term arrangements. The filing said this capacity currently costs significantly more.

The company also recorded a $2 billion (€1.78 billion) non-cash charge linked to a programme matching employee donations. Adjusted operating income reached $2.5 billion (€2.23 billion), representing a 15 per cent margin, down from 19 per cent in the second quarter.

Anthropic’s net loss widened to $23.3 billion (€20.8 billion), compared with $12.1 billion (€10.8 billion) a year earlier. During the first nine months of the year, it recorded losses of $104 billion (€92.9 billion) against revenue of $33.2 billion (€29.6 billion). The company said $94.3 billion (€84.2 billion) of the losses resulted from non-cash accounting charges as the value of convertible notes increased.

The prospectus also disclosed approximately $895.4 billion (€800 billion) in purchase and lease commitments, largely related to computing infrastructure. The commitments are almost seven times the $131.8 billion (€117 billion) Anthropic held in cash and short-term investments at the end of September.

See also  Samsung Enters Smart Ring Market with Galaxy Ring, Aiming to Compete Without Subscription Fees

Around $70 billion (€62.5 billion) is due within 12 months, while approximately $450 billion (€402 billion) falls due by the end of 2030. A further $214 billion (€191 billion) is cancellable or subject to conditions.

The company warned that customers breaching its usage rules or commercial terms accounted for between 5 and 7 per cent of revenue during the first nine months. It said enforcement measures could reduce revenue and future growth.

Anthropic also disclosed potential risks from rapid advances in artificial intelligence, saying it might soon reach its own threshold for dramatically accelerated progress in which AI systems help develop more capable models. The company warned that such developments could pose significant safety concerns, although it has not classified any existing model as having reached that threshold.

The updated filing comes as Anthropic prepares to list on Nasdaq under the ticker ANT. It has not yet announced a share price range, but securities rules require publication of a prospectus at least 15 days before a company begins its investor roadshow.

Anthropic could therefore begin marketing its shares as early as next week, with a listing possible from the week of November 16.

Investor expectations for the company’s valuation have reportedly eased to around $1.5 trillion (€1.33 trillion), down from earlier estimates of approximately $2 trillion (€1.78 trillion), following a technology stock sell-off linked to concerns about AI investment costs. The revised figure would still exceed the $965 billion valuation set during its private funding round in May.

Continue Reading

Tech

Argentina’s Milei Courts AI Data Centre Investors With Tax Incentives and Light Regulation

Published

on

Argentina’s President Javier Milei is seeking to turn Patagonia into a major destination for artificial intelligence data centres, offering international technology companies tax incentives and a promise of minimal regulation as he competes for billions of dollars in investment.

While communities in parts of the United States and Europe are questioning the environmental and electricity costs of large data centres, Milei’s government is actively encouraging their development. The president told the US Chamber of Commerce in Washington last month that Argentina would not regulate artificial intelligence, saying: “We are going to leave it free.”

The country’s southern Patagonia region offers several advantages for energy-intensive facilities, including large areas of available land, cooler temperatures, strong winds and access to natural gas from the Vaca Muerta shale formation.

According to visitor records obtained by The Associated Press, Milei and his aides have held meetings with technology industry figures, including venture capitalist Peter Thiel and representatives of OpenAI, Amazon, Google and Oracle.

Polish renewable energy company Green Capital plans to build a $3 billion data centre in Patagonia, which would provide computing capacity to major technology companies. The project has yet to secure a customer. The company has leased 130,000 hectares of land from 14 local landowners.

Green Capital’s Marta Zlotkowska said Patagonia had emerged as the preferred location after the company compared potential sites worldwide, citing access to land and energy resources.

Tax incentives and questions over benefits

Milei is also promoting a scheme that would reduce corporate income tax from 25% to 15% for investments of at least $1 billion in artificial intelligence and other emerging industries. The proposal is awaiting approval by Argentina’s Senate.

See also  Europe Develops 348-Satellite Network to Reduce Dependence on Starlink

However, critics question how much local communities would gain from the projects. Data centres typically employ relatively small workforces, while the proposed incentives do not require companies to reserve computing capacity for Argentine users or disclose their energy and water consumption.

Environmental concerns are particularly strong near Trelew, a city close to Green Capital’s proposed site. Residents say they have limited information about the project’s water requirements, energy sources and potential impact on local resources. Environmental assessments remain underway and permits have not yet been secured.

If operated continuously at its projected peak capacity, the planned facility would consume around 2.7 times the electricity used by Buenos Aires throughout 2025, according to an AP calculation. Green Capital says it intends to generate its own power through wind, solar and gas-fired facilities.

Opposition grows over regulatory safeguards

Some technology experts warn that Milei’s opposition to AI regulation could leave the country exposed to risks involving surveillance, cyberattacks and personal data. Officials have also acknowledged uncertainty over rules governing copyrighted material used to train AI systems and the transfer of personal information abroad.

In Trelew, residents have organised a campaign called “Trelew Free of Data Centres”, reflecting growing concern over the environmental consequences.

Despite the opposition, some local landowners see the projects as an opportunity to diversify income as drought, rising costs and declining wool production put pressure on traditional sheep farming.

Continue Reading

Tech

Binance Investigates $6.3 Billion Crypto Theft as North Korean Hackers Suspected

Published

on

Cryptocurrency exchange Binance is investigating a security breach that resulted in an estimated $6.3 billion in stolen digital assets, in what could become the largest cryptocurrency theft on record. Early findings point to a compromised employee laptop and possible involvement by a North Korean hacking group.

The incident unfolded over less than four hours, with attackers gaining access to several of Binance’s online wallets and transferring funds to addresses under their control.

Binance, the world’s largest cryptocurrency exchange, said its initial investigation suggested that a compromised laptop belonging to a DevOps engineer may have provided an entry point. The attackers allegedly used access to a service container to submit unauthorised withdrawal transactions.

The company later stressed that the breach involved multiple factors and was more complex than initially understood. It said there was no evidence that wallet private keys had been exposed.

Binance temporarily suspended deposits and withdrawals while carrying out security checks. It also revoked authorisation for the affected addresses and began working with law enforcement agencies and blockchain intelligence companies to trace and recover the assets.

Billions in assets frozen

The scale of the breach became clearer on Friday as Binance issued a series of updates. The exchange initially reported that $2.54 billion had been frozen, before raising the figure to $3 billion.

Blockchain data infrastructure company Bitquery estimated that Binance’s withdrawal system had authorised transfers worth approximately $6.3 billion from 17 hot wallets between Thursday night and early Friday morning.

Stablecoin issuers Tether, Circle and Arbitrum, alongside several smaller companies, helped freeze affected assets to prevent further movement. Bitquery estimated that the attackers had exchanged around $729 million in stablecoins before the freezes took effect.

See also  Alibaba Expands European Data Centres and Unveils New AI Chip

Despite the intervention, approximately $3.1 billion remained under the attackers’ control, primarily in Ether and Bitcoin, according to the company’s analysis. The stolen funds were distributed across nine blockchains, with Ethereum, Bitcoin and TRON among the networks most affected.

Binance said it would cover all losses using company funds and had reserved approximately $6.34 billion for the incident. It also said it had begun recovering affected assets for customers.

The exchange pledged to publish a detailed technical report once the investigation is complete and the findings have been verified. It also plans to release an updated proof of reserves after security measures have been completed.

North Korean group suspected

Blockchain intelligence and cybersecurity firm TRM Labs has attributed the attack to TraderTraitor, a hacking group linked to North Korea.

“Based on robust on- and off-chain intelligence, we make this judgment with high confidence,” TRM Labs said.

The group has previously been linked to major cryptocurrency thefts. The FBI identified TraderTraitor as responsible for the $1.5 billion Bybit exchange hack in February 2025.

Binance has not publicly confirmed the identity of the attackers. Its investigation remains ongoing, with law enforcement agencies and blockchain intelligence partners assisting efforts to trace and recover the remaining funds.

The incident has renewed concerns about the security of cryptocurrency exchanges, particularly the risks associated with online wallets and access to internal technology systems.

Continue Reading

Trending