Google Archives | Digital Watch Observatory

Meta avoids social media addiction trial as wider litigation continues

Meta has avoided a scheduled social media addiction trial after the teenage plaintiff withdrew his claims less than a week before proceedings were due to begin, although broader litigation against major technology platforms over alleged harms to young users continues.

The case was brought by a 15-year-old Florida plaintiff identified as R.K.C. and was expected to become the second bellwether trial examining claims that major social media platforms used allegedly addictive design features that harmed teenagers.

TikTok, Snap and YouTube had already settled the plaintiff’s claims for undisclosed amounts, while his lawyers said he decided to withdraw the remaining case against Meta after weighing the overall outcome of the litigation and the burden of a lengthy trial.

Meta did not reach a settlement. Company spokesperson Andy Stone described the claims as baseless and said Meta would continue defending itself against similar lawsuits.

Litigation against major social media companies continues on several fronts. In an earlier bellwether case, a jury found Meta and Google’s YouTube negligent and awarded one plaintiff a total of US$6 million in compensatory and punitive damages.

In New Mexico, Meta was ordered to pay US$375 million, while further proceedings seeking structural changes to the company’s business practices remain pending.

Seven additional bellwether cases are scheduled in California state court, alongside separate federal litigation in Oakland and lawsuits brought by state attorneys general alleging that Meta misled the public about harmful and allegedly addictive platform features.

Why does it matter?

Although this individual case will not proceed to trial, it forms part of a much wider wave of litigation examining whether social media platforms can be held legally responsible for allegedly addictive design features and their effects on young users. Courts across the United States are increasingly being asked to assess where platform responsibility begins and how companies should balance user engagement with safety.

The remaining bellwether cases could influence future litigation and regulatory debates on child safety, platform accountability and product design. Their outcomes may also shape how courts evaluate claims involving algorithmic recommendation systems and other features designed to maximise user engagement.

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

EU fines Google €890 million under Digital Markets Act

The European Commission has fined Alphabet’s Google €890 million for breaching the Digital Markets Act (DMA), marking one of the bloc’s most significant enforcement actions under its new competition framework for digital platforms.

The Commission imposed a €460 million fine after finding that Google favoured its own services in search results, including shopping, hotels, transport and sports, contrary to the Digital Markets Act (DMA) requirements for fair treatment of competing services.

A separate €430 million fine addressed restrictions within Google Play that prevented app developers from directing users to potentially cheaper offers available through alternative app stores or external websites.

Google has been given 60 days to comply with the European Commission’s decision by treating competing services more fairly and allowing developers greater freedom to direct users outside Google Play.

The company criticised the ruling and indicated it may challenge the decision before the EU courts, arguing that the required changes could reduce the usefulness of Search features and weaken security protections on Google Play.

The Commission nevertheless noted progress in Google’s broader DMA compliance efforts, including ongoing tests that modify how its shopping, hotel and flight services appear in search results, alongside changes affecting advertising, sports and other content.

The Commission also indicated that the principles established by the decision could extend to Google’s AI-powered services, including AI Overviews and AI Mode, signalling that DMA obligations will apply not only to traditional search results but also to emerging generative AI interfaces.

Why does it matter?

The decision represents one of the clearest demonstrations yet of how the Digital Markets Act is being enforced in practice. Rather than focusing solely on financial penalties, the Commission is requiring structural changes to how large digital platforms present services, interact with business users and compete with rivals.

The reference to AI Overviews and AI Mode also suggests that the DMA will increasingly shape the design of AI-powered search services. As generative AI becomes more deeply integrated into online platforms, regulators appear determined to ensure that new interfaces remain subject to the same competition principles as traditional digital services.

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

Google launches Gemini 3.6 Flash for more efficient AI agents

Google has launched Gemini 3.6 Flash, describing it as a faster and more efficient model for developers building AI agents, coding tools and enterprise workflows while reducing the cost of deploying AI at scale.

According to Google, the model uses 17% fewer output tokens than Gemini 3.5 Flash on the Artificial Analysis Index and requires fewer reasoning steps and tool calls for multi-stage tasks, reducing the cost of running production AI agents.

Google has priced Gemini 3.6 Flash at US$1.50 per million input tokens and US$7.50 per million output tokens. The company says it improves coding, computer use, document analysis, chart interpretation and report drafting while reducing unnecessary edits and execution loops.

The company also reported gains on benchmarks including DeepSWE and OSWorld-Verified, alongside stronger safeguards against cyber-offence and chemical, biological, radiological and nuclear misuse.

Google also introduced Gemini 3.5 Flash-Lite, its fastest and lowest-cost model in the 3.5 family, targeting high-volume workloads such as search, document processing and data extraction.

In addition, a specialised Gemini 3.5 Flash Cyber model will power Google’s CodeMender security agent to detect, validate and patch software vulnerabilities. Owing to its potential for misuse, access will initially be restricted to governments and trusted partners through a pilot programme.

Gemini 3.6 Flash and Flash-Lite are available through the Gemini API, Google AI Studio, Android Studio, Gemini Enterprise and the Gemini app, with Flash-Lite also rolling out in Google Search.

Google said Gemini 3.5 Pro remains in partner testing while work is already underway on what it describes as its most ambitious pre-training run yet for Gemini 4.

Why does it matter?

The launch reflects a broader shift in AI development from simply increasing model size towards improving efficiency, reliability and cost-effectiveness for real-world deployment. As AI agents become more widely adopted, reducing inference costs and improving task execution are becoming increasingly important competitive advantages.

The restricted release of Gemini’s cybersecurity model also illustrates how AI developers are adopting more differentiated access policies for high-risk capabilities. Rather than making every model broadly available, companies are increasingly limiting advanced cyber tools to trusted users while attempting to balance defensive benefits with concerns about misuse.

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

European Commission orders Google to open Android and Search under DMA

The European Commission has issued two legally binding specification decisions requiring Google to improve interoperability for competing AI assistants on Android and to share anonymised Google Search data with eligible third-party search providers.

The decisions clarify how Google must comply with its obligations under the Digital Markets Act (DMA) and are intended to strengthen competition in AI assistant and search markets.

Under the Android decision, competing AI assistants will gain access to functions currently available primarily to Google’s own services, including Gemini. European users will be able to activate alternative assistants through voice commands, perform actions within apps, receive suggested replies and ask questions based on recent activity.

The Commission said the measures include safeguards to protect privacy, security and device integrity.

The second decision requires Google to make anonymised search data available under clearer and more effective conditions. Eligible recipients will include AI chatbots offering search functionality, while Google must share the same categories of anonymised data it uses to improve its own search services.

The framework also establishes a multi-layered anonymisation process, allows Google to address serious cybersecurity and data protection risks, and introduces transparent procedures for data access and pricing.

Google must begin sharing search data with eligible providers from January 2027, while the Android interoperability measures are expected to benefit users from July 2027. Although the decisions are legally binding, they do not determine whether Google has breached the DMA or impose financial penalties. They remain subject to judicial review.

Why does it matter?

The decisions represent one of the clearest examples so far of how the Digital Markets Act is intended to reshape competition in digital ecosystems. By requiring Google to open key Android features and search data to rivals, the Commission is seeking to reduce barriers for competing AI assistants and search providers while expanding consumer choice.

The measures also demonstrate that DMA enforcement extends beyond preventing anti-competitive conduct to prescribing how gatekeepers must implement interoperability and data-sharing obligations in practice. Similar specification decisions could shape how other major digital platforms comply with the Act in the future.

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

Google open-sources k8s-aibom to detect shadow AI

Google has open-sourced k8s-aibom, a lightweight Kubernetes controller designed to detect unregistered AI workloads and generate standardised inventories of the AI models, runtimes and frameworks operating inside a cluster.

The tool targets shadow AI: workloads deployed by developers without formal registration or integration with an organisation’s security and governance systems. Such deployments can evade conventional security scanners, particularly where organisations avoid privileged agents, kernel-level access or manual changes to Kubernetes workloads.

Google says k8s-aibom addresses that gap by continuously monitoring Kubernetes APIs and container environments. It detects running AI components and generates CycloneDX 1.6 Machine Learning Bills of Materials (ML-BOMs) based on what is actually executing, rather than what was intended during the build process.

The controller runs as a single unprivileged deployment in the k8s-aibom-system namespace. It does not require sidecars, eBPF modules, privileged DaemonSets or modifications to developers’ continuous integration and deployment pipelines.

The controller monitors KServe resources, deployments, StatefulSets, DaemonSets and jobs across a cluster. It then analyses container images, environment variables and command-line arguments to identify different categories of AI workloads.

Supported systems include inference runtimes such as vLLM, Triton Inference Server, TGI, and Ollama; agent frameworks including LangChain, AutoGen, and CrewAI; retrieval and vector database tools such as Milvus, Qdrant, and pgvector; and distributed training and evaluation workloads.

Once identified, the components are compiled into CycloneDX ML-BOM documents. These records can be stored as Kubernetes custom resources or exported to destinations including Google Cloud Storage and webhook endpoints.

Google also designed the tool to produce identical ML-BOM documents when given identical cluster inputs. This deterministic behaviour is intended to support GitOps workflows, allowing security and reliability teams to compare records and identify changes when AI dependencies drift.

Unlike build-time scanners, which document what organisations intended to deploy, k8s-aibom observes live clusters to identify which AI systems are actually running, how they are connected and how those findings were established.

A confidence model separates detected components into three categories. Declared assets are explicitly specified in workload configurations, inferred assets are identified through runtime patterns, and unresolved assets indicate that an AI presence was detected but the precise model, version, or weights could not be established.

Unresolved findings can therefore be prioritised for further security review, while declared and inferred classifications help auditors distinguish documented engineering intent from conclusions reached by the controller.

Google says the controller follows least-privilege principles and can export records using a dedicated identity with permission to create objects in Cloud Storage. Creation preconditions can prevent existing ML-BOM records from being silently overwritten, strengthening the historical evidence available to security and compliance teams.

Google also positions k8s-aibom as a tool for regulatory and standards compliance. Runtime inventories could help organisations gather evidence relevant to the EU AI Act, the NIST AI Risk Management Framework and ISO/IEC 42001 requirements for AI asset management.

Why does it matter?

Shadow AI has become a growing governance challenge as developers deploy AI tools outside formal security and compliance processes. Without visibility into what is actually running in production, organisations may struggle to assess risk, investigate incidents or demonstrate regulatory compliance.

By generating inventories of live AI workloads rather than relying solely on build-time records, k8s-aibom could help organisations improve AI governance while supporting audits, security operations and compliance with emerging AI standards and regulations.

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

UK CMA consults on Apple and Google app store payment rules

The UK Competition and Markets Authority has opened consultations on new requirements for Apple and Google under the country’s digital markets competition regime.

Proposed steering requirements would allow app developers to direct UK users to payment options outside Apple’s App Store and Google’s Play Store. The CMA said Apple currently bans steering in the UK, while Google restricts it.

According to the regulator, allowing developers to communicate with customers about off-platform options could increase competition, reduce payment costs and support innovation across mobile services.

Consultation proposals also cover principles to ensure that any steering fees charged by Apple and Google are fair and reasonable. The CMA said such fees should be based on evidence and should be lower than current app store charges.

Alongside the steering proposals, officials are seeking views on a potential requirement for Apple to provide developers with access to near-field communication functionality on iOS.

Broader NFC access could allow UK fintechs and developers to support contactless payments from within their own apps. It could also support future payment methods, including account-to-account payments, digital currencies and stablecoins, as well as non-financial uses such as digital ID and car keys.

Responses to the steering conduct requirement are due by 28 July 2026, while views on the potential NFC requirement are due by 21 July 2026. The CMA will decide later this year whether to impose new obligations.

Why does it matter?

The consultations show the UK’s digital markets regime moving into targeted conduct rules for major mobile platforms. If adopted, the measures could weaken Apple and Google’s control over in-app payments and give developers more freedom to offer alternative purchasing channels. The NFC proposal also widens the debate beyond app store commissions, addressing Apple’s control over device functionality that can shape competition in mobile payments, digital identity and other services.

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

Google proposes a balanced approach to AI governance in the US

Google has published a policy paper proposing a two-track approach to AI governance in the United States, separating oversight of frontier AI models from rules for widely deployed AI applications.

The paper argues that AI policy should avoid what Google describes as a false choice between over-regulation and no regulation. Instead, the company calls for a pragmatic, evidence-based framework that treats the most advanced AI systems differently from everyday AI tools such as chatbots.

For frontier AI, Google proposes the creation of a Frontier AI Regulatory Organisation, or FARO. The industry-funded body would operate under federal oversight and develop standards for safety, security, incident reporting and transparency.

Google says FARO could set scientific benchmarks for frontier capabilities, particularly in areas such as cybersecurity and chemical, biological, radiological and nuclear risks. It could also oversee independent audits and require frontier AI companies to publish and follow safety frameworks before releasing highly capable models.

For widely deployed AI applications, Google argues that the federal government should rely mainly on existing legal frameworks, with targeted updates where needed. The paper says policy should focus on real-world harms and outputs rather than micromanaging AI development.

The company identifies several priority areas, including workforce preparedness, child safety, information integrity, copyright, privacy and energy infrastructure for data centres.

Google supports measures such as AI interaction guidelines for children, disclosures that chatbots are not sentient, rules for self-harm-related queries, watermarking and provenance standards for generative AI, privacy-enhancing technologies and workforce reskilling.

The paper presents the model as a way to address national security and consumer protection risks while preserving US leadership in AI development.

Why does it matter?

Google’s paper is a significant industry intervention in the US AI policy debate. Its two-track model reflects a broader governance trend: frontier AI is increasingly being treated as a national security and safety issue, while everyday AI applications are being handled through consumer protection, child safety, privacy, copyright and labour policy. The proposal could influence federal discussions, but it also reflects Google’s own regulatory preferences, including industry-funded oversight, confidential audit reports and reliance on existing law for many AI applications.

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

Google launches Gemini for Science AI research tools

Google has introduced Gemini for Science, a collection of AI experiments and tools designed to support scientific discovery across research fields.

The initiative includes three experimental tools on Google Labs. Hypothesis Generation, built with Co-Scientist, helps researchers define research challenges, generate hypotheses and evaluate them through a multi-agent process. Google said the tool uses an ‘idea tournament’ in which agents generate, debate and assess possible research directions, with claims supported by clickable citations.

Computational Discovery, built with AlphaEvolve and Empirical Research Assistance, is designed to generate and score large numbers of code variations in parallel. Google said the prototype could help scientists test modelling approaches in areas such as solar forecasting and epidemiology.

Literature Insights, built with NotebookLM, searches scientific literature and organises results into structured tables for side-by-side analysis. Researchers can use it to identify research gaps, synthesise findings across papers and create outputs such as reports, slide decks and audio or video overviews.

Google said access to the experiments will open gradually through Google Labs. The company is also bringing related capabilities to enterprise organisations through Google Cloud, with partners testing tools for pharmaceutical research, crop science, supply chain optimisation and work linked to the US Department of Energy’s Genesis Mission.

As part of Gemini for Science, Google is also launching Science Skills, a bundle that integrates more than 30 life science databases and tools, including UniProt, the AlphaFold Database, AlphaGenome API and InterPro. Google said the tools can support workflows such as structural bioinformatics and genomic analysis on agentic platforms such as Google Antigravity.

The company said it is working with more than 100 institutions to validate its scientific AI systems and has created a trusted tester community that includes PhD students, industry researchers and Nobel laureates.

The launch shows how major AI developers are moving from specialised scientific models towards broader agentic tools that support hypothesis generation, literature analysis and computational testing.

Why does it matter?

Gemini for Science points to a wider shift in AI-assisted research: AI systems are moving beyond literature search or single-task modelling towards multi-step scientific workflows. Such tools help researchers navigate large bodies of literature, test computational ideas faster and identify new hypotheses. But their value will depend on evidence quality, reproducibility, peer review and clear limits around what AI-generated scientific suggestions can and cannot prove.

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

Google expands financial ad verification across EU and EEA

Google has announced the expansion of its financial services advertiser verification programme to every country in the EU and European Economic Area, extending requirements aimed at reducing fraudulent financial advertising.

The rollout will cover 24 additional countries and builds on an existing programme already active in six EU member states and the United Kingdom.

Under the programme, advertisers seeking to promote financial products or services must complete an additional verification process showing that the relevant national regulator authorises them. Google said it will check credentials against official registries across the EU and EEA.

The requirements will be introduced in phases. Businesses will have 30 days to complete the process after notification, and unverified advertisers will have their financial services ads restricted until verification is completed.

Google said the additional requirements build on its wider advertiser identity verification programme, which it says already covers more than 98% of ads seen across the EU. The company also said its systems blocked or removed more than 1.6 billion ads in the EU last year.

The expansion comes amid continuing concern over online financial scams, including fraudulent ads that impersonate legitimate financial services providers or promote misleading investment products.

Why does it matter?

Financial scams increasingly rely on digital advertising to reach consumers at scale. Google’s expansion adds another gatekeeping layer for financial advertisers across Europe by linking ad eligibility to authorisation in official regulatory registers. The measure also shows how large platforms are being pushed, by regulators and reputational pressure, to take more responsibility for the trustworthiness of high-risk advertising categories such as finance.

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

Los Angeles AI arts museum Dataland opens with Google Cloud support

Dataland, a Los Angeles museum dedicated to AI-based art, has opened to the public with Google serving as a technology and creative collaborator.

The museum was co-founded by media artist Refik Anadol and Efsun Erkılıç and is located at The Grand LA in downtown Los Angeles. Google says the 25,000-square-foot space is designed as an interactive environment where data, machine learning and sensory experiences form part of the artwork.

Its inaugural exhibition, ‘Machine Dreams: Rainforest’, uses Anadol’s Large Nature Model, an AI system trained on environmental datasets, to transform natural-world data into large-scale generative visuals.

Google Cloud provides infrastructure for the museum’s real-time image generation, soundscapes, scent augmentation and interactive visitor experiences. Google says the system uses tools including Gemini, diffusion models and generative adversarial networks.

The project builds on a decade of collaboration between Google and Anadol, including work using LA Philharmonic archives, Google Quantum AI data, planetary datasets and the ‘Machine Dreams: Biophilia’ installation at Google’s Mountain View campus.

Google Arts & Culture is also supporting the Dataland AI Artist Residency, a six-month programme for four artists. The residency will provide grants, mentorship from Refik Anadol Studio and access to Google Cloud tools and machine learning models.

Why does it matter?

Dataland shows how AI art is moving from experimental installations into permanent cultural infrastructure. It also highlights the role of cloud providers and large AI platforms in shaping creative production, exhibition design and access to machine-learning tools. For cultural institutions, the project raises broader questions about authorship, data provenance, sustainability, audience interaction and the dependence of new creative formats on private technology infrastructure.

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