Economic Archives | Page 284 of 685 | Digital Watch Observatory

Eric Schmidt warns that AI growth is limited by electricity

Former Google chief executive Eric Schmidt has warned that electricity, rather than semiconductors, will limit the future growth of AI.

Speaking on the Moonshots podcast, Schmidt said the push towards artificial superintelligence—AI that exceeds human cognitive ability in almost all domains—will depend on securing sufficient power instead of just developing more advanced chips.

Schmidt noted the US alone may require an extra 92 gigawatts of electricity to support AI growth, equivalent to dozens of nuclear power stations.

Instead of waiting for new plants, companies such as Microsoft are seeking to retrofit closed facilities, including the Three Mile Island plant targeted for relaunch in 2028.

Schmidt highlighted growing environmental pressures, citing Microsoft’s 34% increase in water use within a year, a trend experts link directly to rising AI workloads.

Major AI developers like OpenAI’s Sam Altman also acknowledge energy as a key constraint. Altman has invested in nuclear fusion through Helion, while firms such as Microsoft and AMD are pressing US policymakers to fast-track energy permits.

Environmental groups, including Greenpeace, warn that unchecked AI expansion risks undermining climate goals instead of supporting them.

Schmidt believes superintelligence is inevitable and approaching rapidly, predicting specialised AI tools across all fields within five years. Rather than focusing solely on AI’s capabilities, he stressed the urgent need for planning energy infrastructure today to match tomorrow’s AI demands.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Amazon cuts hundreds of AWS jobs amid AI shift

Amazon has reportedly cut hundreds of jobs within its Amazon Web Services cloud computing division, according to sources familiar with the matter.

Instead of providing specific numbers, Amazon confirmed the decision, citing ongoing efforts to optimise resources while continuing to invest in customer-focused innovation.

These layoffs follow warnings by CEO Andy Jassy, who recently said the rising use of generative AI would reduce workforce needs across the company.

The cuts affect several AWS teams, including specialists who support customers in developing new products and selling services. Employees reported receiving termination emails on Thursday, with their system access deactivated soon after.

Rather than focusing solely on AWS, Amazon has also been reducing roles in other units, such as its books division, devices and services unit, and Wondery podcast platform.

Despite these workforce reductions, AWS sales rose 17% in the first quarter to $29.3 billion compared to a year earlier, with operating income increasing by 23% to $11.5 billion.

Amazon, alongside firms like Microsoft, Meta, and CrowdStrike, is increasingly relying on AI tools instead of human workers to automate tasks, write software code, and streamline operations as part of a broader trend affecting the tech industry.

Amazon’s latest cuts reflect efforts by Jassy to reduce bureaucracy and eliminate managerial layers, aiming to keep the company agile amid rapid AI adoption and changing business priorities.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Irish hospital turns to AI for appointment management

Beaumont Hospital in Dublin plans to deploy AI to predict patient no-shows and late cancellations, aiming to reduce wasted resources.

Instead of relying solely on reminders, the hospital will pilot AI software costing up to €110,000, using patient data to forecast missed appointments. Currently, no-shows account for 15.5% of its outpatient slots.

The system will integrate with Beaumont’s existing two-way text messaging service. Rather than sending uniform reminders, the AI model will tailor messages based on the likelihood of attendance while providing hospital staff with real-time insights to better manage clinic schedules.

The pilot is expected to begin in late 2025 or early 2026, potentially expanding into a full €1.2 million contract.

The move forms part of Beaumont Hospital’s strategic plan through 2030 to reduce outpatient non-attendance. It follows the broader adoption of AI in Irish healthcare, including Mater Hospital’s recent launch of an AI and Digital Health centre designed to tackle clinical challenges using new technologies.

Instead of viewing AI as a future option, Irish hospitals now increasingly treat it as an immediate solution to operational inefficiencies, hoping it will transform healthcare delivery and improve patient service.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Sberbank prepares to offer cryptocurrency custody services

Russia’s largest lender, Sberbank, is preparing to offer custody services for cryptocurrency assets, providing security and safeguards against hacking. The bank has proposed treating Bitcoin and digital assets like traditional bank accounts to Russia’s central bank.

These proposals include measures to freeze assets if law enforcement suspects illegal activity.

The central bank has eased its stance, supporting digital assets for international trade to bypass Western sanctions over the Ukraine conflict. Sberbank’s executive director highlighted the growing global trend of banks offering custody services and expressed a desire for Russia to keep pace.

In addition to custody, Sberbank has launched Bitcoin-linked structured bonds and plans to roll out Bitcoin futures and similar investment products on the Moscow Exchange.

Industry experts emphasise that establishing local custody services is vital to reduce reliance on foreign companies and strengthen the security of Russia’s crypto market.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Mistral’s chatbot Le Chat takes on ChatGPT with major upgrade

France-based AI startup Mistral has rolled out a major update to Le Chat, its AI chatbot, introducing new features aimed at challenging rivals like ChatGPT, Gemini and Claude. The update includes Deep Research, voice interaction, reasoning capabilities and a refreshed image editor.

According to the company’s latest blog post, the new Deep Research mode transforms Le Chat into a structured assistant that can clarify needs, search sources and deliver summarised findings. The tool enables users to receive comprehensive responses in a neatly formatted report.

In addition, Mistral unveiled Vocal mode, allowing users to speak to the chatbot as if they were talking to a person. The feature is powered by the firm’s voice input model, Voxtral, which handles voice recognition in real time.

The company also introduced Think mode, based on its Magistral reasoning model. Designed for multilingual and complex tasks, the feature provides thoughtful and clear responses, even when answering legal or professional queries in languages like Spanish or Japanese.

For users juggling multiple conversations or tasks, the new Projects tool groups related chats into separate spaces. Each project includes a dedicated Library for storing files and content, while also remembering individual tools and settings.

Users can upload documents directly into Projects and revisit past chats or references. Content from the Library can also be pulled into the active conversation, supporting a more seamless and personalised experience.

A revamped image editor rounds out the update, offering users the ability to tweak AI-generated visuals while maintaining consistency in character design and fine details. Mistral says the upgrade helps improve image customisation without compromising visual integrity.

All features are now available through Le Chat’s web platform at ‘chat.mistral.ai’ or via the company’s mobile apps on Android and iOS. The update reflects Mistral’s growing ambition to differentiate itself in the increasingly competitive AI assistant market.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Hackers hide malware using DNS TXT records

Hackers are increasingly exploiting DNS records to deliver malware undetected, according to new research from DomainTools.

Instead of relying on typical delivery methods such as emails or downloads, attackers now hide malicious code within DNS TXT records, part of the Domain Name System, often overlooked by security systems.

The method involves converting malware into hexadecimal code, splitting it into small segments, and storing each chunk in the TXT record of subdomains under domains like whitetreecollective.com.

Once attackers gain limited access to a network, they retrieve these chunks via ordinary-looking DNS queries, reassembling them into functioning malware without triggering antivirus or firewall alerts.

The rising use of encrypted DNS protocols like DNS-over-HTTPS and DNS-over-TLS makes detecting such queries harder, especially without in-house DNS resolvers equipped for deep inspection.

Researchers also noted that attackers are using DNS TXT records for malware and embedding harmful text designed to manipulate AI systems through prompt injection.

Ian Campbell of DomainTools warns that even organisations with strong security measures struggle to detect such DNS-based threats due to the hidden nature of the traffic.

Instead of focusing solely on traditional defences, organisations are advised to monitor DNS traffic closely, log and inspect queries through internal resolvers, and restrict DNS access to trusted sources. Educating teams on these emerging threats remains essential for maintaining robust cybersecurity.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Stay True To The Act campaign defends music rights

More than 30 European musicians have launched a united video campaign urging the European Commission to preserve the integrity of the EU AI Act.

The Stay True To The Act campaign calls on policymakers to enforce transparency and uphold copyright protections.

Artists, including Spanish singer-songwriter Álex Ubago and Poland’s Eurovision 2025 entrant Justyna Steczkowska, have voiced concern over the unauthorised use of their work to train AI models. They demand the right to be informed and the power to refuse such usage.

The EU AI Act, passed in 2024, includes provisions requiring developers to disclose the content used in AI training. However, as implementation plans develop, artists fear the law may be diluted, weakening protections for creators.

The campaign appeals for vigorous enforcement of the Act’s original principles: transparency, copyright control and fair innovation. Artists say AI and music can coexist in Europe only if ethical boundaries are upheld.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

OpenAI releases ChatGPT agent with task automation tools

OpenAI has introduced a general-purpose AI agent within ChatGPT, aiming to move beyond answering questions by completing various computer-based tasks.

Known as ChatGPT agent, the tool allows users to navigate calendars, generate editable presentations, and write code simply by using natural language prompts.

Rather than acting as a standalone product, ChatGPT agent integrates capabilities from OpenAI’s earlier tools, combining website navigation and in-depth research features.

Rolling out to Pro, Plus, and Team subscribers, the ChatGPT agent also connects with external apps like Gmail and GitHub. Instead of being limited to basic queries, it can access a terminal and use APIs, enabling tasks such as analysing competitors or planning shopping lists.

OpenAI claims its underlying model delivers state-of-the-art results, scoring significantly higher than previous versions on academic and maths benchmarks.

While positioning ChatGPT as its most capable AI tool yet, OpenAI has implemented several new safety measures due to the agent’s potential risks. The company acknowledges its model could amplify harm in sensitive areas like biological and chemical threats.

To mitigate such dangers, OpenAI monitors prompts in real time and turns off ChatGPT’s memory feature within the agent to avoid data leaks through malicious attacks.

Despite these precautions, questions remain over whether the ChatGPT agent will consistently perform complex tasks in the real world. Earlier agent technologies from various companies have often failed to meet expectations.

OpenAI, however, insists its new release represents a more robust step towards fulfilling the vision of practical AI agents.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

xAI eyes data centre deal with Humain

Elon Musk’s AI venture, xAI, has entered early discussions with Humain to secure data centre capacity instead of relying solely on existing infrastructure.

According to Bloomberg, the arrangement could involve several gigawatts of capacity, although Humain has yet to start building its facilities, meaning any deal would take years to materialise.

Humain is backed by Saudi Arabia’s Crown Prince Mohammed bin Salman and the Public Investment Fund (PIF). xAI is reportedly considering a fresh funding round where PIF might also invest.

At the same time, xAI is negotiating with a smaller company constructing a 200-megawatt data centre, offering a more immediate solution while waiting for larger projects.

Rather than operating in isolation, xAI joins AI competitors like Google, Meta and Microsoft in racing to secure vast computing power for training large AI models. The push for massive data centre capacity reflects the escalating demands of advanced AI systems.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Quantum sector surges as Rigetti unveils modular machine

Quantum computing is progressing faster than expected, with Rigetti Computing achieving a major breakthrough. The company reached 99.5% median 2-qubit gate fidelity using a modular 36-qubit system—halving its error rate and moving closer to fault-tolerant computing.

The new machine, built from four 9-qubit chiplets, represents the industry’s largest multichip quantum system. Its modular design addresses the scalability challenges of single-chip models.

Rigetti’s superconducting qubits also outperform rivals by operating at significantly faster speeds.

Plans are in place to launch a 100+-qubit system before the end of 2025. Backing from DARPA, the US Air Force, and the UK government further validates Rigetti’s approach. Partnerships with Nvidia and Quanta Computer add commercial strength.

Despite modest revenue, Rigetti holds $575 million in cash with no debt and owns the sector’s first dedicated quantum chip factory. True commercial quantum advantage is expected between 2026 and 2028.

Shares of Rigetti, along with other quantum computing firms like IonQ and D-Wave, have surged in recent months, outperforming the broader market.

Rigetti offers strong potential—but remains a high-risk pick in a competitive field.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!