Sustainable development

Google acknowledges AI being used for harmful content

Google has reported receiving over 250 complaints globally about its AI software being used to create deepfake terrorist content, according to Australia’s eSafety Commission.

The tech giant also acknowledged dozens of user reports alleging that its AI program, Gemini, was being exploited to generate child abuse material. Under Australian law, companies must provide regular updates on their efforts to minimise harm or risk hefty fines.

The eSafety Commission described Google’s disclosure as a ‘world-first insight’ into how AI tools may be misused to produce harmful and illegal content.

Between April 2023 and February 2024, Google received 258 reports of suspected AI-generated extremist material and 86 related to child exploitation. However, the company did not specify how many of these reports were verified.

A Google spokesperson stated that the company strictly prohibits AI-generated content related to terrorism, child abuse, and other illegal activities.

While it uses automated detection to remove AI-generated child exploitation material, the same system is not applied to extremist content.

Meanwhile, the regulator has previously fined platforms like X (formerly Twitter) and Telegram for failing to meet reporting requirements, with both companies planning to appeal.

For more information on these topics, visit diplomacy.edu.

Singapore expands charges in server fraud case

Singapore authorities have filed additional charges against three men in a widening investigation into server fraud, which may involve AI chips, court documents revealed on Thursday.

The suspects are accused of deceiving tech firms Dell and Super Micro by falsely representing the final destination of the servers they purchased.

Officials have stated the servers could contain Nvidia chips but have not confirmed whether they fall under US export controls.

The case is part of a broader probe involving 22 individuals and companies suspected of fraudulent transactions. US authorities are also investigating whether Chinese AI firm DeepSeek has been using restricted American chips.

Singapore has confirmed that some servers were sent to Malaysia, where authorities are now examining if any laws were violated.

Two suspects, Aaron Woon and Alan Wei, face additional fraud charges, while a third, Li Ming, had his earlier charge updated to include an alleged offence dating back to 2023.

Lawyers representing the men have either declined to comment or stated that the case is complex due to its international scope.

Meanwhile, Singapore police have seized 42 electronic devices and are analysing bank statements as they work with foreign law enforcement to trace the movement of funds.

For more information on these topics, visit diplomacy.edu.

Microsoft drops parts of CoreWeave partnership

Microsoft has reportedly scaled back parts of its agreements with cloud computing provider CoreWeave due to delivery issues and missed deadlines, according to the Financial Times.

Despite maintaining several contracts worth billions, the tech giant has moved away from certain deals, though the decision is said to be unrelated to any broader changes in its data centre strategy.

CoreWeave, backed by Nvidia, specialises in providing high-powered AI computing resources and competes with major cloud providers like Microsoft Azure and Amazon AWS.

The company is preparing for a major initial public offering (IPO) in New York, aiming for a valuation exceeding $35 billion and seeking to raise over $3 billion.

The cloud provider recently expanded its capabilities by acquiring AI developer platform Weights & Biases for an undisclosed sum.

Neither Microsoft, CoreWeave, nor Nvidia have responded to requests for comment regarding the report.

For more information on these topics, visit diplomacy.edu.

Alibaba unveils new AI model boosting shares by 8%

Alibaba Group’s latest AI breakthrough sent its Hong Kong-listed shares soaring more than 8% on Thursday.

The e-commerce giant announced its new AI reasoning model, QwQ-32B, which it claims performs on par with DeepSeek’s R1 model despite having far fewer parameters—32 billion compared to 671 billion.

The unveiling comes as China pushes for advancements in AI, humanoid robots, and 6G technology, with government support pledged on Wednesday.

Alibaba’s new model is available through its chatbot service, Qwen Chat, which offers various AI models, including the high-powered Qwen2.5-Max.

The company said QwQ-32B demonstrated strong performance in areas such as mathematical reasoning, coding, and problem-solving, competing closely with top global models like OpenAI’s o1 mini and DeepSeek’s R1.

DeepSeek has gained recognition in China for producing highly efficient AI models at a fraction of the cost of its Western rivals.

Meanwhile, another Chinese AI innovation also made headlines on Thursday. A startup called Monica introduced an AI agent named Manus, which reportedly outperformed OpenAI’s Deep Research on an AI assistant benchmark.

Unlike traditional chatbots, AI agents are designed for more complex tasks, with Manus capable of creating travel itineraries and comparing insurance policies.

Currently available by invitation only, the agent gained significant attention online, with a demonstration video amassing over 280,000 views and users eager to gain access.

For more information on these topics, visit diplomacy.edu.

Google unveils experimental AI search for premium users

Google has introduced an experimental version of its search engine that removes the traditional 10 blue links in favour of AI-generated summaries.

The new ‘AI Mode’ is available to subscribers of Google One AI Premium, a $19.99 per month plan, and can be accessed through a tab alongside existing options like Images and Maps.

Users will see a detailed AI summary with hyperlinks to cited sources, replacing standard search results with a search bar for follow-up questions.

The feature is powered by a customised version of Google’s Gemini 2.0 model, designed to handle complex queries more effectively.

AI Overviews, which provide summaries atop search results, are already available in over 100 countries, with advertisements integrated into them since last May. Google says the new AI-driven approach responds to demand from “power users” seeking more AI-generated responses.

As Google pushes deeper into AI-powered search, it faces competition from Microsoft-backed OpenAI, which introduced search capabilities to ChatGPT last October.

The shift has raised concerns among content creators, with edtech company Chegg suing Google in February, alleging that AI previews are reducing demand for original content and hurting publishers’ ability to compete.

For more information on these topics, visit diplomacy.edu.

Scale AI wins Pentagon AI contract

Scale AI has secured a multimillion-dollar contract with the US Department of Defense to develop AI tools for military operations. The project, named Thunderforge, is the Pentagon’s flagship AI initiative aimed at enhancing decision-making, simulation, and operational planning. Led by the Defense Innovation Unit, the programme will see collaboration with technology partners such as Microsoft and Anduril, initially focusing on the Indo-Pacific and European Commands before expanding further.

Thunderforge represents a shift towards AI-driven military strategies, promising increased speed and efficiency in modern warfare. The Department of Defense and Scale AI have emphasised the importance of rapid response times, with the technology designed to process vast amounts of information quickly. However, while Scale AI has assured that human oversight will be maintained, the Defence Innovation Unit has not explicitly highlighted this aspect in its communications.

The deal comes as more AI firms reconsider their stance on military applications. Companies like OpenAI, Microsoft, and Google have altered policies that previously restricted AI development for defence purposes. Critics argue that these technologies could be used for harmful applications despite company assurances, raising ethical concerns about the potential for AI in warfare. Some experts warn that firms may have limited control over how their AI is ultimately deployed.

This latest partnership highlights the growing integration of AI into defence strategies, despite previous pushback from tech industry employees. While firms insist their technologies will be used responsibly, concerns remain over the long-term implications of AI-driven warfare and the ability to ensure its ethical use.

For more information on these topics, visit diplomacy.edu.

Nvidia shares drop after US tariffs announcement

Nvidia’s shares plummeted nearly 9% on 3 March following an announcement by US President Donald Trump confirming new tariffs on imports from Canada and Mexico, set to take effect on 4 March. The decline contributed to a broader market downturn, with the Dow Jones falling by 800 points and the Nasdaq dropping by over 3%. Nvidia’s market value took a sharp hit, losing around $265 billion and falling to $2.79 trillion, a steep drop from its previous $3 trillion valuation.

Despite reporting strong earnings, with revenue surging 78% year-over-year to $39.33 billion, Nvidia’s stock has lost 13% since 26 February. The 25% tariffs could affect the company’s operations, particularly as some of its systems are manufactured in the US and Mexico. However, CEO Jensen Huang remains optimistic, highlighting Nvidia’s AI advancements and the upcoming Blackwell chips, which he says will drive strong performance in the next quarter.

Nvidia also plans to play a key role in Taiwan Semiconductor’s $100 billion expansion in the US, a project mentioned by Trump. While the company faces short-term market volatility and policy challenges, its long-term strategy remains focused on technological growth and innovation.

For more information on these topics, visit diplomacy.edu.

EU set to implement Entry/Exit System for border control

The European Union has reached an agreement to finally implement the long-awaited Entry/Exit System (EES), which will modernise border checks for short-stay travellers.

After several delays due to technical issues and a lack of readiness, the system is now set to begin operations in autumn, though a specific launch date has yet to be determined.

Member states will have the option to introduce the system all at once or in phases over a six-month transition period, with the full implementation to be completed by the end of the transition.

The EES, which was first proposed in 2016, aims to replace traditional passport stamping by collecting biometric data from non-EU visitors, including photos and fingerprints.

This data will be recorded each time visitors enter or exit the Schengen Area. The system is designed to improve border control, help authorities identify overstayers, and prevent identity fraud.

While Cyprus and Ireland will not participate in the new system, all other EU member states and four Schengen-associated countries will be involved.

Poland, which currently holds the EU Council’s rotating presidency, will lead negotiations with the European Parliament to finalise the law.

Tomasz Siemoniak, Poland’s Minister for Internal Affairs, indicated that a final agreement should be reached smoothly, with October set as the target for full implementation.

The EES is expected to provide authorities with new tools to enhance security and better manage borders within the Schengen Area.

For more information on these topics, visit diplomacy.edu.

South Korea launches $34 billion fund for strategic industries

South Korea has announced the creation of a $34 billion policy fund to support companies in key industries such as semiconductors, automotive, and advanced technologies, in response to growing global competition and protectionist policies.

The state-run Korea Development Bank will manage the fund by providing low-interest loans and other financial support over the next five years to businesses involved in national strategic industries.

The government stressed that maintaining competitiveness in these strategic sectors has become crucial to the country’s economic security, particularly amid the uncertainties caused by the new US administration.

South Korea has identified 12 industries, including semiconductors, AI, and biopharmaceuticals, as critical for its future economic stability and will offer targeted financial support to strengthen these sectors.

In addition to the fund, South Korea also unveiled new policies to attract skilled global talent in cutting-edge fields. These measures include offering top-tier visas and permanent residency to professionals with experience at major international firms, aiming to enhance the country’s workforce in strategic industries.

For more information on these topics, visit diplomacy.edu.

Musk’s bid to halt OpenAI’s for-profit transition rejected

A US court has denied Elon Musk’s request for a preliminary injunction against OpenAI’s transition into a for-profit organisation.

US District Judge Yvonne Gonzalez Rogers ruled that Musk did not meet the high standards required to block the move. However, the judge indicated that she would expedite a trial on the matter, which is expected to take place later this year.

Musk, who co-founded OpenAI in 2015 but left before its major success, argued that OpenAI had initially sought his charitable funding to create AI for the public good, but has since shifted its focus towards making profits.

His lawyer, Marc Toberoff, expressed satisfaction that the judge had agreed to a swift trial, claiming that the case involves urgent public interest concerns.

OpenAI, which is seeking to become a for-profit entity to attract the necessary capital for its AI projects, welcomed the court’s decision.

The company emphasised that its goal is to develop advanced AI models to benefit society. Musk’s legal action, which also includes antitrust claims, stems from his frustration with OpenAI’s shift in direction since he departed from the organisation.

For more information on these topics, visit diplomacy.edu.