Access Archives | Page 11 of 329 | Digital Watch Observatory

New iPhone vulnerability raises concerns over advanced mobile cyber threats

A newly identified cyberattack known as ‘DarkSword’ is raising concerns about the security of iPhone devices, following reports that millions of users could be exposed to rapid data extraction techniques.

Cybersecurity researchers indicate that the attack targets specific iOS versions, exploiting vulnerabilities in the Safari browser and a graphics processing feature known as WebGPU.

Once access is gained, attackers can retrieve sensitive information, including messages, emails and location data, within minutes, while removing traces of the intrusion almost immediately.

Estimates suggest that a significant share of global iPhone users may be affected, with hundreds of millions of devices running vulnerable software versions.

The scale of exposure remains uncertain, particularly as experts continue to assess whether additional versions of iOS may also be impacted.

Researchers have associated the campaign with a threat actor previously identified by Google, with observed activity across multiple regions.

Such a development highlights growing concerns about the evolution of mobile cyber threats, where increasingly sophisticated techniques are being deployed beyond traditional state-level operations.

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

TikTok disinformation study raises concerns over AI content and EU regulation

A new study by Science Feedback indicates that TikTok has a higher proportion of misleading content than other major platforms operating in the EU.

The analysis covered France, Poland, Slovakia and Spain, assessing content across multiple thematic areas including health, politics and climate.

Findings suggest that approximately one in four posts on TikTok contained misleading elements, placing the platform ahead of competitors such as Facebook, YouTube and X. Health-related narratives were the most prominent category, reflecting broader patterns observed across digital ecosystems.

Researchers describe disinformation as a persistent feature embedded within platform structures instead of an isolated occurrence.

The study also highlights a growing presence of AI-generated content, particularly in video formats, where synthetic material accounted for a significant share of misleading posts. Despite existing platform policies, most identified content lacked clear labelling.

The regulatory context remains under development.

While the Digital Services Act integrates voluntary commitments from the EU disinformation code, it does not impose mandatory requirements for identifying AI-generated material.

Ongoing debates therefore focus on transparency, accountability and the evolving responsibilities of digital platforms within the European information environment.

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

EU scrutiny intensifies over Broadcom VMware licensing dispute

Broadcom is facing increased regulatory pressure in the EU following a formal antitrust complaint concerning changes to VMware licensing practices.

The complaint highlights growing tensions between large technology providers and European cloud infrastructure firms.

The filing, submitted by Cloud Infrastructure Services Providers in Europe, raises concerns that revised licensing models could significantly alter market dynamics.

European providers argue that the changes may limit flexibility, increase costs, and affect their ability to compete effectively in the cloud services sector.

At the centre of the dispute lies the broader issue of market concentration and control over critical digital infrastructure.

Industry stakeholders suggest that restrictive licensing conditions could reshape access to essential virtualisation technologies, which underpin a wide range of cloud and enterprise services across the EU.

Regulatory attention is expected to focus on whether such practices align with the EU competition rules, particularly regarding fair access and market neutrality.

The case emerges at a time when European policymakers are intensifying oversight of dominant technology firms and seeking to strengthen digital sovereignty across strategic sectors.

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

UK drops AI copyright opt-out plan amid growing industry divide

The UK Government has abandoned its previous preference for an AI copyright opt-out model, signalling a shift in policy following strong opposition from creative industries.

Ministers now acknowledge that there is no clear consensus on how AI developers should access copyrighted material.

Concerns from writers, artists and rights holders focused on the use of their work in training AI systems without permission.

Liz Kendall confirmed that extensive consultation exposed significant disagreement, prompting the government to step back from its earlier position that would have allowed the use of copyrighted content unless creators opted out.

A joint report from the Department for Science, Innovation and Technology and the Department for Culture, Media and Sport states that further evidence is required before any legislative change.

Policymakers in the UK will assess how copyright frameworks influence AI development, while also examining international regulation, licensing models and ongoing legal disputes.

Government strategy now centres on balancing innovation with fair compensation.

Officials emphasise that creators must retain control over how their work is used, while AI developers require access to high-quality data to remain competitive. Potential measures include labelling AI-generated content to reduce risks linked to disinformation and deepfakes.

No timeline has been set for reform, reflecting the complexity of aligning economic growth with intellectual property protection.

The debate unfolds alongside broader ambitions outlined by Rachel Reeves, who has identified AI as a central driver of future economic expansion, with the UK aiming to lead adoption across the G7.

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

AI safety push sees Anthropic and OpenAI recruit explosives specialists

Anthropic and OpenAI are recruiting chemical and explosives experts to strengthen safeguards for their AI systems, reflecting growing concern about the potential misuse of advanced models.

Anthropic is seeking a policy specialist to design and monitor guardrails governing how its systems respond to prompts involving chemical weapons and explosives. The role includes assessing high-risk scenarios and responding to potential escalation signals in real time.

OpenAI is expanding its Preparedness team, hiring researchers and a threat modeller to identify and forecast risks linked to frontier AI systems. The positions focus on evaluating catastrophic risks and aligning technical, policy, and governance responses.

The recruitment drive comes amid heightened scrutiny of AI safety and national security implications. Anthropic is currently challenging a US government designation that labels it a supply-chain risk, while tensions have emerged over restrictions on the military use of AI systems.

At the same time, OpenAI has secured agreements to deploy its technology in classified environments under defined constraints. The parallel developments highlight how AI firms are balancing commercial expansion with increasing pressure to implement robust safety controls.

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

Data centres drive LG’s integrated AI infrastructure push

AI infrastructure is becoming a central battleground for growth, with LG Group accelerating its push into AI data centres and energy storage systems under its ‘One LG’ strategy.

The initiative brings together key affiliates to deliver integrated solutions for AI data centres. LG Electronics provides cooling systems, LG Energy Solution handles power infrastructure, including ESS and UPS, while LG Uplus and LG CNS oversee design, construction, and operations.

The strategy comes as global demand for AI data centres surges, driven by energy-intensive workloads and rising electricity constraints. Expanding storage capacity has become critical, with the US expected to add over 24 gigawatts of energy storage capacity in 2026 alone.

LG Electronics is focusing on advanced cooling technologies, including large air-cooled chillers and liquid-cooling systems, to manage the intense heat generated by GPU-intensive AI workloads. The company has also expanded into immersion cooling through partnerships, aiming to achieve efficiency gains in next-generation facilities.

Meanwhile, LG Energy Solution is strengthening its role in power infrastructure, scaling ESS production across North America, and securing major contracts. Through integrated battery and software solutions, the company is positioning itself to meet growing demand for stable, high-capacity energy systems supporting AI operations.

On the networking side, LG Uplus is developing low-latency infrastructure and AI-driven data centre management systems to optimise performance and energy use in real time. Together, these efforts highlight LG’s ambition to become a full-stack provider in the rapidly expanding AI data centre ecosystem.

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

NVIDIA Isaac powers generalist specialist robots at scale

A new class of robots is emerging, combining broad adaptability with task-specific precision as developers move toward generalist specialist systems. Within this shift, NVIDIA Isaac is enabling integrated workflows that connect data generation, simulation, training, and deployment across robotics pipelines.

NVIDIA Isaac unifies robotics development across these stages, integrating cloud-to-robot workflows that allow developers to build, test, and scale systems more efficiently across both real and simulated environments.

A key driver is the growing reliance on synthetic data, which allows developers to simulate rare or hazardous scenarios that are difficult to capture in the real world. NVIDIA Isaac supports this through tools such as Omniverse-based simulation and teleoperation pipelines, helping convert real-world signals into scalable training datasets and accelerating development cycles.

The platform also enables advanced robot training using reasoning vision-language-action models, which allow machines to perceive, interpret, and act across complex environments. With frameworks like Isaac Lab and integrated physics engines, NVIDIA Isaac enables robots to train across thousands of parallel simulations, significantly reducing time, cost, and risk compared to real-world training.

Once trained, NVIDIA Isaac supports deployment across edge AI systems, including the Jetson platform, while maintaining consistency between simulation and real-world performance. Combined with modular workflows and open frameworks, the platform is positioning itself as a core foundation for scalable, next-generation robotics.

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

Quantum cryptography pioneers win top computing prize

Two researchers have been awarded the Turing Award for pioneering work in quantum cryptography. Their research laid the foundations for a new form of secure communication based on quantum physics.

The method, developed in the 1980s, enables encryption keys that cannot be copied without detection. Any attempt to intercept the data alters its physical properties, revealing interference.

Experts say the approach could become vital as quantum computing advances. Traditional encryption methods may become vulnerable as computing power increases.

The award highlights the growing importance of secure data transmission in a digital world. Researchers believe quantum cryptography could play a central role in encrypting and protecting future communications.

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

EU advances AI simplification effort ahead of further negotiations

A committee within the European Parliament has approved a proposal to simplify aspects of AI regulation, marking a step forward in efforts to refine the implementation of the AI Act.

An initiative that seeks to adjust certain requirements to support clearer compliance, particularly for industry stakeholders.

The proposal focuses on technical and procedural elements linked to how AI rules are applied in practice.

Lawmakers aim to ensure that regulatory obligations remain proportionate while maintaining existing safeguards. Part of the discussion includes how specific categories of AI systems should be addressed within the broader framework.

Some elements of the proposal may require further discussion in upcoming negotiations with the Council of the European Union. Areas under consideration include the treatment of sensitive AI applications and the balance between regulatory clarity and enforcement effectiveness.

The development reflects ongoing efforts within the EU to refine its approach to AI governance. As discussions continue, policymakers are expected to assess how adjustments can support innovation while maintaining consistency with existing legal principles.

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

Meta to end Instagram private message encryption after May 8

The US tech giant, Meta, has announced that end-to-end encryption for private messages on Instagram will no longer be supported after 8 May.

Previously, such a technology ensured that only intended recipients could read messages, preventing even Meta from accessing their contents.

The decision follows concerns from law enforcement and child protection organisations, which argued that encrypted messages can make it harder to identify harmful content involving children.

Meta has stated that the update allows the platform to monitor messages while maintaining standard privacy safeguards.

End-to-end encryption had been the default for several messaging platforms, including WhatsApp, Messenger, and other Meta services.

The company first signalled its intent to expand encryption across Instagram and Messenger in 2019, implementing it in 2023. The plan was met with objections from organisations such as the Internet Watch Foundation and the Virtual Global Taskforce.

These groups highlighted potential risks in preventing the timely detection of harmful content, particularly child sexual abuse material.

Meta’s shift reflects a compromise between privacy, platform security, and online child safety. The company has not provided further details on changes to encryption policies beyond Instagram’s private messaging service.

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