United News of Bangladesh

A high-profile legal battle between Elon Musk and Sam Altman has intensified scrutiny on OpenAI’s leadership, with Altman seen as having the most at stake as he prepares to testify in court. The civil trial, underway in Oakland, California, stems from a lawsuit filed by Musk accusing Altman of abandoning the founding mission of OpenAI— originally established as a nonprofit—to pursue a profit-driven model. Musk is seeking Altman’s removal from leadership and financial support for the company’s charitable arm. Testimonies over the past two weeks have highlighted internal tensions at OpenAI, including Altman’s brief removal as CEO in 2023. A text exchange between Altman and company executive Mira Murati during that period—where she described the situation as “very bad”—has drawn widespread attention online. Witnesses, including former board members Helen Toner and Tasha McCauley, raised concerns about Altman’s leadership, citing what they described as a “pattern of behavior” involving lack of transparency and resistance to board oversight. OpenAI co-founder Ilya Sutskever also testified, saying Altman’s conduct had affected trust and productivity within the company. However, he later supported Altman’s reinstatement, citing fears that the company could collapse without him. Despite the criticism, OpenAI’s current board chair Bret Taylor defended Altman, describing his performance as CEO as effective and transparent. Experts say the trial could have wider implications for the artificial intelligence sector, particularly as OpenAI, Musk’s ventures, and rival firms move toward potential initial public offerings. While Musk also faces risks, including scrutiny tied to his business interests, analysts suggest Altman’s leadership future could hinge on his courtroom testimony. “This is not looking good for any of them,” said Sarah Kreps of Cornell University, noting that the case comes at a time when public perception of AI is already under strain. The jury is expected to weigh conflicting accounts of leadership and governance before delivering a verdict that could significantly impact one of the world’s most influential AI companies.  

An experimental café in Sweden’s capital is drawing attention for putting an artificial intelligence system in charge of daily operations, while human baristas continue to serve customers. The café, launched by San Francisco-based startup Andon Labs, is being managed by an AI agent named “Mona,” powered by Google Gemini. The system handles key business functions such as hiring, inventory management and operational decisions, leaving staff to prepare and serve food and drinks. Since opening in mid-April in Stockholm, the café has generated over $5,700 in sales, though expenses have significantly reduced its initial budget of more than $21,000. Much of the spending went toward setup costs, and operators hope the business will eventually become profitable. Customers have shown curiosity about the AI-run concept, with some interacting directly with the system via a phone inside the café. “It’s nice to see what happens if you push the boundary,” said one visitor, adding that the quality of the drinks remained good. Experts worry about AI's role going forward Researchers and analysts say the experiment raises important ethical and practical questions, particularly around accountability and decision-making. Emrah Karakaya, an academic at KTH Royal Institute of Technology, warned that handing control to AI could create risks if proper oversight mechanisms are not in place. “If something goes wrong, such as a food safety issue, it raises questions about responsibility,” he said, cautioning that poorly managed systems could harm businesses and society. Founded in 2023, Andon Labs focuses on testing AI systems in real-world scenarios using actual financial resources and operational tools. The company has collaborated with major AI developers, including OpenAI, Anthropic and Google DeepMind. Company officials say the Stockholm café is part of a broader effort to understand how AI might run organizations autonomously in the future. AI agent struggles with inventory orders While the AI system has successfully handled tasks such as securing permits, setting up services and recruiting staff through platforms like LinkedIn and Indeed, it has faced challenges in day-to-day management. The system has reportedly over-ordered supplies — including thousands of napkins and gloves — and purchased items not used in the café. It has also struggled with managing bakery deliveries, occasionally ordering too much or missing deadlines, forcing menu changes. Developers attribute these issues to limitations in the AI’s memory and context processing. Despite the experiment, café staff say their roles remain secure for now, suggesting that AI may pose a greater challenge to managerial positions than to frontline workers.  

A human barista serves coffee, but an artificial intelligence agent is running operations at an experimental café in Stockholm, where almost all business decisions are handled by AI in a real-world test of automation. The Andon Café, operated by San Francisco-based startup Andon Labs, has placed an AI agent named “Mona” in charge of management tasks while human staff continue to prepare and serve drinks. Powered by Google’s Gemini, the system oversees hiring, contracts, inventory and day-to-day operations. International cyberattack disrupts universities and schools across three countries Since opening in mid-April, the café has earned over $5,700 in sales but remains below break-even, with less than $5,000 left from an initial budget exceeding $21,000. Organisers say most early spending went on setup costs and expect performance to stabilise over time. Customers have reacted with curiosity, with visitors able to call the AI via an in-café telephone. “It’s nice to see what happens if you push the boundary. The drink was good,” said customer Kajsa Norin. Andon Labs, founded in 2023, says it is stress-testing AI systems in real environments and has worked with OpenAI, Anthropic, Google DeepMind and xAI. It envisions AI-run organisations in the future. Technical staff member Hanna Petersson said the café is a controlled experiment to study ethical questions in AI-led management. However, experts including KTH’s Emrah Karakaya warned of accountability risks, calling it “opening Pandora’s box.” Mona has made major errors, including over-ordering supplies and mismanaging bread deliveries. Barista Kajetan Grzelczak said jobs remain safe for now, but middle management could be affected.

A widespread cyberattack has disrupted universities and schools in the United States, Canada and Australia, causing major academic interruptions during a critical end-of-semester period. The hacking group ShinyHunters has claimed responsibility for the attack, which temporarily knocked offline Canvas, a widely used online academic platform operated by Instructure and used by thousands of educational institutions worldwide. By Thursday night, Instructure said Canvas had been restored “for most users,” though several universities continued to report outages on Friday. The company estimated that around 9,000 institutions globally were affected. The disruption forced multiple universities to postpone or cancel examinations and coursework. Mississippi State University, for example, delayed final exams to allow students to recover lost academic work. Students described confusion and panic after a ransom message appeared on screens mid-exam, claiming that ShinyHunters had breached the system and threatening to release stolen data unless a payment in cryptocurrency was made. Universities including Penn State University, Idaho State University and the University of British Columbia confirmed service interruptions, with some cancelling exams and warning students not to attempt logging into the platform. The Chicago Maroon, the university-led newspaper, posted a screenshot of a message from ShinyHunters that appeared to be seeking a ransom. The message encouraged the university to contact the hacking group privately "to negotiate a settlement" and avoiding "the release of their data". It was the same message that Northwestern University masters student Jacques Abou-Rizk said he received when he clicked a link in an email that appeared to be from a university administrator. "I didn't know what was happening," Abou-Rizk recalled. "It's a scary message to receive." At the University of Toronto, officials said multiple institutions were impacted, while students at the University of California, Los Angeles faced difficulties submitting assignments. The University of Chicago temporarily disabled its Canvas access following the breach. The hacking group reportedly demanded ransom payments in exchange for not leaking stolen data. Some students and faculty described widespread confusion as systems suddenly went offline during exams and assignments. Cybersecurity analysts said the attack appeared to be part of a coordinated extortion campaign, with deadlines issued for payment demands earlier this week. The incident comes amid growing concerns about cyber threats targeting educational and public infrastructure, with US lawmakers urging stronger national cybersecurity measures. Authorities and Instructure continue to investigate the breach, while universities are working to restore services and reschedule affected academic activities. Source: BBC  

China has unveiled a set of guidelines aimed at regulating and advancing the innovative development of artificial intelligence (AI) agents, as the country accelerates efforts under its “AI plus” initiative. The guidelines, jointly issued by the Cyberspace Administration of China, the National Development and Reform Commission(NDRC) and the Ministry of Industry and Information Technology(MIIT), are intended to support the implementation of the State Council’s “AI plus” action plan and promote the orderly growth of intelligent agents as key forms of AI products and services, according to an official statement. Instagram drops private messaging feature worldwide The document defines AI agents as intelligent systems capable of autonomous perception, memory, decision-making, interaction and execution, noting that they are increasingly integrating with cyberspace and the physical world with the advancement of technologies such as large language models. It outlines key principles for AI agent development, emphasizing safety and controllability,  standardization and order, innovation-driven growth, and application-oriented development. The guidelines identify four major areas of action. These include strengthening development foundations by improving technological infrastructure and establishing standards and protocols; ensuring safety and security; promoting application-driven development through 19 typical scenarios across scientific research, industrial development, consumption, public services and social governance; and fostering an innovation ecosystem through enhanced industrial collaboration and broader application promotion.

Instagram users around the world can no longer send highly private direct messages after parent company Meta disabled end-to-end encryption (E2EE) on the platform from Thursday. The move marks a major shift by Meta, which had earlier promoted the technology as the future standard for online privacy. End-to-end encryption is considered one of the safest forms of digital communication because only the sender and receiver can read messages. Critics, however, have long argued that it can also make it harder for authorities to detect crimes such as child abuse and online grooming. With the feature now removed, Instagram will be able to access the contents of direct messages, including photos, videos and voice notes. Meta had announced plans in 2019 to expand E2EE across Facebook and Instagram messaging services. While Facebook Messenger fully adopted the system in 2023, Instagram only offered it as an optional feature and had planned to make it standard later. The company has now abandoned those plans after seven years. Instead, Instagram will use standard encryption, a system commonly used by services such as Gmail, where providers can access data if necessary. Child safety organisations welcomed the decision. The NSPCC, a UK-based children’s charity, said the removal of E2EE would help prevent criminals from hiding abusive activities online. “We are really pleased,” said Rani Govender of the NSPCC, adding that encrypted messaging could allow child abuse and grooming to go undetected. Privacy campaigners strongly criticised the decision, calling it a setback for online security. Maya Thomas of privacy group Big Brother Watch said E2EE helps protect children’s personal data and warned that Meta may be responding to government pressure. Meta did not publicly announce the policy reversal but quietly updated Instagram’s terms and conditions in March, stating that end-to-end encrypted messaging would no longer be supported after May 8, 2026. The company said the decision was made because only a small number of users had enabled the feature. Cybersecurity experts, however, believe the move may also reflect Meta’s growing interest in using data for targeted advertising and artificial intelligence development. Victoria Baines, professor of information technology at Gresham College, said social media companies increasingly see user communications as valuable data for AI training and advertising purposes. Instagram previously said direct messages were not used to train AI systems. The decision could also affect the wider social media industry, where encrypted messaging had increasingly become common. Apps such as WhatsApp, Signal, Facebook Messenger, iMessage and Google Messages currently use E2EE by default. Meanwhile, Telegram offers the feature as optional, while TikTok recently said it has no plans to introduce end-to-end encryption for direct messages. Experts say Meta’s latest move could slow the wider expansion of encrypted messaging on social media platforms. With inputs from BBC.

Concerns over the risks posed by artificial intelligence have emerged as a major theme in the ongoing courtroom battle between billionaire Elon Musk and Sam Altman, even though the technology itself is not formally on trial. The federal trial in Oakland, California, centres on Musk’s lawsuit accusing Altman and fellow OpenAI leaders of abandoning the organisation’s original nonprofit mission. Musk claims the company betrayed its founding promise to develop artificial intelligence for the benefit of humanity rather than private profit. OpenAI, however, argues that Musk is attempting to slow down the company to help his own AI venture, xAI. Although Judge Yvonne Gonzalez Rogers warned lawyers not to turn the case into a debate over AI safety, discussions about the technology’s dangers have repeatedly surfaced during testimony. Witnesses and lawyers have raised concerns ranging from job losses and misinformation to discrimination and emotional dependency on AI chatbots. Some testimony also touched on Musk’s long-standing warnings that highly advanced AI could eventually threaten humanity itself. AI researcher Stuart Russell testified as an expert witness for Musk’s legal team, saying the intense competition among companies to dominate artificial intelligence could itself pose risks. Russell, a computer scientist at the University of California, Berkeley, told the court that whichever company first develops artificial general intelligence (AGI) could gain enormous power over competitors. He listed several concerns linked to AI, including racial and gender bias, job displacement, misinformation and psychological harm to some chatbot users. “Whichever company develops AGI first would have a very big advantage,” Russell told jurors. The trial focuses on the founding of OpenAI in 2015 as a nonprofit research organisation backed in part by Musk. Both Musk and Altman have publicly stated that OpenAI was created to safely develop advanced AI for humanity’s benefit, though each side now accuses the other of trying to gain control over the technology. A nine-member jury from the San Francisco Bay Area will decide whose claims are more credible. Before testimony began, Judge Gonzalez Rogers cautioned lawyers against broad discussions about whether AI is harmful to humanity. “This is not a trial on the safety risks of artificial intelligence. This is not a trial on whether or not AI has damaged humanity,” the judge said. Despite that warning, Musk used part of his testimony last week to discuss his concerns about the future of AI. He described AGI as a stage where AI becomes “as smart as any human” and claimed society is approaching that point rapidly. “We are getting close to that point,” Musk said, adding that AI could surpass human intelligence as early as next year. Musk said he had long viewed AI as “a double-edged sword” and was worried about the concentration of power within major technology firms such as Google. According to Musk, one of the reasons for helping establish OpenAI was to create a counterbalance to Google’s dominance in AI development. During testimony, Musk also said he intentionally helped create OpenAI as a nonprofit organisation rather than a profit-driven company. “I deliberately chose this for the public good,” he said. The judge questioned that argument, noting that Musk later launched his own AI company, xAI, which now operates in the same field. OpenAI co-founder and president Greg Brockman defended the organisation’s mission during his testimony this week. Brockman said OpenAI’s work was always intended to serve humanity rather than any single corporation or individual. “It was about humanity as a whole,” he said. He also claimed Musk sought personal control over the company in its early days. According to Brockman, Musk initially appeared supportive of Altman becoming OpenAI’s chief executive but later insisted people needed to know “he was in charge.” Musk is seeking damages in the case and also wants Altman removed from OpenAI’s board. Legal experts say a victory for Musk could complicate OpenAI’s plans for a future stock market listing.

South Korea’s KOSPI surged nearly 7% to an all-time high on Wednesday, led by a sharp rally in tech shares as optimism grows over artificial intelligence expansion and easing geopolitical tensions. Shares of Samsung Electronics jumped almost 13%, while SK Hynix rose about 10% in early trading. Both firms are key producers of advanced chips essential for AI technologies. Investor sentiment improved further following reports that Iranian officials were heading to China ahead of a planned meeting between Donald Trump and Xi Jinping. The development also helped stabilize oil market fluctuations. After remaining closed Tuesday for a public holiday, South Korea’s market reopened strongly, with the KOSPI climbing 6.7% early in the session to 7,398.34. Elsewhere in Asia, stock markets mostly moved higher, though trading in Tokyo was shut due to a holiday. Australia’s S&P/ASX 200 advanced nearly 1% to 8,766.80, while Hong Kong’s Hang Seng rose 0.7% to 26,081.52. China’s Shanghai Composite also gained 1% to 4,152.68. In energy markets, benchmark U.S. crude fell $1.37 to $100.90 per barrel, and Brent crude dropped $1.50 to $108.37. Despite recent declines, prices remain significantly higher than pre-war levels near $70. U.S. military officials said a ceasefire with Iran is in place, though uncertainty persists. Efforts are ongoing to secure shipping routes through the Strait of Hormuz, a critical passage for global oil transport. On Wall Street, the S&P 500 rose 0.8% to a record close of 7,259.22. The Dow Jones Industrial Average gained 0.7% to 49,298.25, while the Nasdaq Composite climbed 1% to a new high of 25,326.13. U.S. economic data offered mixed signals. One report indicated slower-than-expected growth in the services sector, partly attributed to the war’s impact on spending. Another showed job openings were slightly higher than forecast at the end of March, suggesting continued resilience in the labor market. In currency trading, the U.S. dollar edged down to 157.88 Japanese yen, while the euro strengthened to $1.1720.

Prosecutors in New Mexico have called for sweeping changes to social media platforms owned by Meta as a landmark trial enters its second phase, focusing on child safety concerns linked to apps like Instagram, Facebook and WhatsApp. Opening statements began Monday in a three-week bench trial to determine whether Meta’s platforms constitute a public nuisance. In the first phase, a jury imposed $375 million in civil penalties, finding the company knowingly harmed children’s mental health and failed to address child sexual exploitation risks. State prosecutors are now urging the court to mandate fundamental reforms, including curbing addictive features, strengthening age verification systems and enhancing privacy settings to better protect minors. New Mexico Attorney General Raúl Torrez said the earlier verdict challenged long-standing legal protections for tech firms under Section 230 of the US Communications Decency Act. The state has proposed a $3.7 billion plan aimed at addressing the impact of social media on children’s mental health. Measures include redesigning content recommendation algorithms to reduce engagement-driven exposure, limiting features such as infinite scrolling and push notifications, and introducing safeguards against exploitation. Prosecutors are also seeking requirements for parental oversight of child accounts and the appointment of a court-monitored safety supervisor to track compliance. Meta has opposed the proposals, arguing they are impractical and infringe on free speech and parental rights. The company has warned it could withdraw services from New Mexico if compelled to implement what it describes as excessive regulations. During the hearing, District Court Judge Bryan Biedscheid expressed caution about overstepping judicial authority, while legal experts noted that applying public nuisance laws to internet platforms remains largely untested. Meta’s legal team also questioned the feasibility of implementing state-specific changes to its apps, calling such demands technologically unworkable. The case is the first to go to trial among dozens filed by US states accusing Meta of contributing to a youth mental health crisis. Its outcome could have far-reaching implications for how social media platforms operate and are regulated across the country.

Asian stock markets mostly declined on Tuesday, following losses on Wall Street where U.S. equities pulled back from recent record highs. Oil prices, which had surged earlier due to rising tensions between the United States and Iran, retreated slightly. Meanwhile, U.S. stock futures posted modest gains of about 0.1%. Trading activity across the region remained subdued, with markets in Japan, South Korea and mainland China closed for public holidays. In Hong Kong, the Hang Seng Index dropped 1.3% to 25,757.56, while Taiwan’s Taiex index rose 0.3%. Australia’s S&P/ASX 200 fell 0.4% to 8,659.90 after the Reserve Bank of Australia increased its key interest rate to 4.35% in response to inflation pressures driven by rising fuel and commodity costs linked to Middle East instability. This marked the central bank’s third rate hike of 0.25 percentage points this year. The bank reported Australia’s annual inflation rate stood at 4.6% through March, above its target range of 2% to 3%. Elsewhere, India’s Sensex index declined 0.7%. Geopolitical tensions remained a key concern. A fragile ceasefire between the United States and Iran appeared to weaken after the U.S. military said it had sunk six Iranian boats allegedly targeting civilian vessels. Despite two U.S.-flagged ships passing through the Strait of Hormuz, the crucial route for global oil and gas shipments remains largely restricted. The United States has also imposed a maritime blockade on Iranian ports, while President Donald Trump launched “Project Freedom,” aimed at escorting stranded vessels through the strait. Global oil benchmarks declined after recent gains. Brent crude fell $1.13 to $113.31 per barrel after briefly exceeding $114, while U.S. crude dropped $2.04 to $104.38 per barrel. Prices had been near $70 before the conflict escalated in late February. Efforts to reach a lasting peace have stalled, with tensions rising further after the United Arab Emirates reported an Iranian attack — the first since last month’s ceasefire. Analysts warned that the situation could deteriorate further, noting that ongoing maritime operations may heighten risks in the Persian Gulf region. On Wall Street, the S&P 500 fell 0.4% to 7,200.75, while the Dow Jones Industrial Average dropped 1.1% to 48,941.90. The Nasdaq Composite slipped 0.2% to 25,067.80. Shares of GameStop plunged 10.1% after it announced plans to acquire eBay, a firm with a market value roughly four times larger. In currency trading, the U.S. dollar edged up slightly against the Japanese yen, while the euro weakened marginally against the dollar.