Kyle said London’s large capital markets could help scale up technology firms capable of competing with American and Asian giants. “We are the spin-out capital of Europe. We are the unicorn capital of Europe,” Kyle later told reporters, referring to the creation of new companies and start-ups valued at more than ?bn. He added that Britain wanted to go further and would be more likely to achieve global reach by working with European countries and the European Union. Britain signed a similar, later-suspended, deal with the United States in September to align on innovation and encourage private-sector investment.

The proposed EU technology pact comes amid wider negotiations under the reset in relations promised by British Prime Minister Keir Starmer, who is seeking to stimulate Britain’s weak economy. Kyle met Sefcovic as the EU and UK prepare for a summit at a yet-to-be confirmed date, likely in July. Both sides hope to present agreements on food and animal safety standards, a youth mobility scheme, and the linking of their emissions trading systems, but talks have encountered roadblocks. Britain is said to want a cap on visas under the mobility scheme and to be reluctant to pay into some EU funds sought by Brussels. The EU, on the other hand, has been demanding greater access to British universities for its 18- to 30-year-olds, and for them to be allowed to pay the same tuition fees paid by local peers.

The tool, known as the Eureka system, works alongside robotic or laparoscopic surgery and highlights different parts of the anatomy in real time on a screen. Connective tissue could be highlighted as turquoise or nerves could be highlighted in green, while other parts of the anatomy can be shown with other colours. Experts said the system supports precision, efficiency and safety by helping surgeons protect or dissect specific structures during operations.

The portable Artificial Intelligence unit was developed by surgeons in Japan who trained the tool using thousands of videos of surgical procedures. The operation on Thursday was the first time it has been used in the UK. It was also the first time that it has been used during surgery outside Japan. Consultant surgeon Kapil Sahnan described the tool as an “extra helping arm” that can look at live surgery and identify hidden structures that may not be visible to surgeons.

Sahnan compared the system to a surgical navigation aid, saying its real-time overlay allows surgeons to see guidance while operating. He said the aim is to prevent errors before they happen by using intelligence derived from thousands and thousands and thousands of labelled operative videos. Work is under way to prove the system’s advantages and explore wider rollout, with Sahnan saying it would be “amazing if everybody had it in the next couple of years” because it could make surgery safer.

Doosan Robotics is integrating NVIDIA Isaac Sim, NVIDIA Isaac Lab, NVIDIA Cosmos open world foundation models, the open source Newton physics engine and NVIDIA Jetson Thor into its Agentic Robot OS. The platform is designed to connect perception, reasoning, simulation, learning and on-device inference. By integrating NVIDIA’s physical Artificial Intelligence technologies, Doosan Robotics aims to make industrial robots better able to perceive, reason and act in complex and dynamic environments. Simulation-to-real workflows, physics calibration and Artificial Intelligence reasoning are expected to make collaborative robots more adaptable, task-specialized and ready for scalable deployment.

The companies also plan to develop reference use cases for industrial tasks such as depalletizing and sanding, while exploring new robot form factors including dual-arm and humanoid platforms. Doosan Robotics aims to evolve from a robot arm provider into a full-stack Artificial Intelligence-first robotics solution company. Doosan Bobcat plans to explore NVIDIA physical Artificial Intelligence technologies for equipment used in construction, landscaping, agriculture and material handling, with a focus on specialized world models that help machines perceive environments, reason about changing conditions and perform tasks more autonomously.

Doosan Enerbility is exploring ways to support NVIDIA Artificial Intelligence factories and the NVIDIA DSX Artificial Intelligence factory platform through large-scale power infrastructure, including gas turbines, steam turbines and small modular reactors, alongside Doosan Fuel Cell’s hydrogen fuel-cell systems. Future work could include power supply design for Artificial Intelligence factory deployments, optimization of generation equipment and evaluation of low-carbon power sources. Doosan Corporation Electro-Materials BG is also supporting next-generation Artificial Intelligence data center infrastructure through copper clad laminate, a foundational material for printed circuit boards used in networking equipment, Artificial Intelligence accelerators and Artificial Intelligence server motherboards.

The suit arrives as Congress has not passed a federal Artificial Intelligence safety regime, leaving courts and state enforcers to define the boundaries of responsibility. Users and families have already brought cases against chatbot companies over alleged mental health harms, but Uthmeier is the first attorney general to bring such a claim. Lawyers involved in online safety litigation see state enforcement as a complement to private suits, because attorneys general can act on behalf of the broader public rather than only specific victims.

Key legal defenses remain unsettled. One difference concerns Section 230 of the Communications Decency Act, the 1996 regulation that shields online platforms from liability for user-posted content. Some lower courts have found that the shield does not apply when lawsuits target platform design rather than user posts, and legal experts said chatbot companies may have a weaker claim because the chatbot itself produces the disputed speech. Some social media companies have invoked the 2024 Supreme Court case Moody v. NetChoice, which established that platforms have an expressive right to design algorithms for placing a priority on certain content.

Plaintiffs may still face evidentiary hurdles. Social media cases have relied on studies about mental health effects, while the research base for Artificial Intelligence chatbots is thinner and the models keep changing. Defense arguments may focus on causation and whether alleged harms were reasonably foreseeable. Safety advocates counter that chatbot interactions can make the causal link clearer when a user treats a bot as a therapist or confidant and receives harmful responses shortly before a crisis. Technology industry representatives are seeking national legislation that sets safety rules and guardrails before courts create a patchwork of standards through litigation.

Industry executives see potential benefits for creators and YouTube. Remixing can lower the barrier to production, help fans and creators localize or edit videos more easily, and extend the lifespan and reach of existing content. Newly minted creators could use the tool to produce b-roll and storytelling assets they might not otherwise be able to make. Still, some marketers warn that easier generation could oversaturate the creator economy and weaken the authenticity that makes influencer marketing effective. Concerns also center on the pressure to appear polished online and the risk that Artificial Intelligence-generated work may normalize unrealistic standards for creators.

The strongest objections involve consent, manipulation, copyright and trust. The opt-out process appears to require disabling remixing on each individual Short, raising concerns that creators may miss the setting or fail to give explicit permission. Remixed Shorts link back to original content, but may not send meaningful traffic to the source, and highly altered versions could distort a creator’s tone, identity or sponsor relationships. Lawyers also warn that Gemini-generated material may create uncertainty under YouTube’s copyright rules if third parties make claims. On June 9, New York State’s Synthetic Performer Disclosure Law will take effect, in which Artificial Intelligence “performers” in ads have to be disclosed. Similar laws will take effect in California and the European Union in early August.

Creators and brands remain cautious about fully Artificial Intelligence-generated deliverables. A November 2025 study from Billion Dollar Boy found that 58% of creators are interested in exploring copyright protection for their face, identity and voice. Artificial Intelligence-generated content can also undermine brand identity and run afoul of IP law: 55% of marketers and 53% of creators say that Artificial Intelligence has led to more copyright infringement and IP theft in the creator economy, according to the Billion Dollar Boy study. A recent study from Ipsos and Syracuse University tested 20 ads across 10 major brands, some human-made, others Artificial Intelligence-generated, and found that 38% of participants felt human-made ads were more creative, and 46% felt they were more emotionally engaging.

Battery health remains one of the biggest concerns for electric vehicle buyers, because replacement can be expensive and frequent fast charging can stress battery cells over time. According to data collected by Recharged, most Tesla batteries still keep around 85% to 90% of their original capacity after driving nearly 200,000 miles. Some Tesla vehicles may even reach 300,000 to 400,000 miles before battery capacity drops to 70%.

The researchers used reinforcement learning, a machine learning method in which a system tests actions and improves based on results. Instead of applying the same charging pattern every time, the Artificial Intelligence system studies the battery’s charge level, age, overall health, and charging history before charging begins. It then adjusts the charging current in real time, unlike traditional systems that rely on fixed current and voltage settings regardless of battery condition.

Using the Artificial Intelligence-based charging strategy, researchers achieved a 22.9% increase in battery lifetime measured in equivalent full charging cycles. The average charging time using the Artificial Intelligence system was 24.12 minutes. Traditional charging methods averaged 24.15 minutes. The method is designed to reduce lithium plating, a damaging process that can occur during fast charging when metallic deposits form on the battery surface instead of lithium being stored correctly. The study, titled “Lifelong Reinforcement Learning for Health-Aware Fast Charging of Lithium-Ion Batteries”, was published in IEEE Transactions on Transportation Electrification and led by Changfu Zou and Meng Yuan.

Real-world testing on physical battery packs and vehicles is still needed before automakers can fully adopt the technology. According to Grand View Research, the global EV charging infrastructure market was worth about $40.22bn in 2025. It could grow to $50.2bn in 2026 and eventually reach nearly $239bn by 2033. The market is expected to grow at a compound annual growth rate (CAGR) of 25%. Fast chargers already dominate the market, accounting for more than 73% of charging infrastructure in 2025. Researchers say the charging method could be added through software updates in existing battery management systems, though calibration would be needed for different battery chemistries and vehicle designs.

The plan frames foreign dependence as a strategic vulnerability. It says too much Canadian innovation is captured and scaled elsewhere, while sovereign compute capacity remains nascent and Canadian organisations rely heavily on foreign providers for the infrastructure behind economic, scientific and public-sector activity. GPU chip fabrication sits “almost entirely offshore”, and only 12pc of Canadian businesses currently use Artificial Intelligence, well behind Nordic counterparts, where adoption runs between 29 and 42pc.

The six pillars cover safety and democracy protections, Artificial Intelligence skills and literacy for all Canadians, accelerated adoption across the economy, building sovereign compute infrastructure, scaling Canadian Artificial Intelligence champions, and forging trusted international alliances. On infrastructure, the Canadian government is committing to building a world-leading supercomputer by 2031 and growing sovereign cloud capacity to reduce dependence on foreign providers, echoing the EU’s CADA (Cloud and Artificial Intelligence Development Act) proposals published on Wednesday.

Canada aims to increase business Artificial Intelligence adoption from 12pc today to 60pc by 2034, create up to 250,000 new jobs through Artificial Intelligence adoption by 2031, and create nearly $200bn in GDP gains from labour productivity improvements. Priority sectors for investment will be: health and life sciences; energy and natural resources; transportation; agriculture; and manufacturing and robotics.

The strategy flags that Canada has already signed 20 new economic and defence international partnerships in the past year, 11 of which advance Artificial Intelligence cooperation. The Canadian government said it will build a strategic multilateral alliance to move “from reliance to resilience” in key Artificial Intelligence and technology capabilities. The plan also commits to modernising privacy legislation, introducing online safety laws and providing free Artificial Intelligence literacy training to 1m entry-level, post-secondary students.

Darius Osborne, a Howard University graduate who could not secure an entry-level newsroom role, became a data labelling analyst for Meta while continuing to freelance. His work involves reviewing Artificial Intelligence-generated videos, prompts or hooks and labelling whether they meet quality standards, including clarity, visual artifacts, surreal-looking errors or inappropriate content. In Canada, Khaleda Khan spent four to five months searching for entry-level media work after graduating in late 2024 before joining xAI as a trainer for the company’s Artificial Intelligence chatbot, Grok. For the six months Khan was at xAI, her job centered on improving Artificial Intelligence systems through a mix of transcription, linguistic review, and quality control, or being an editor of a robot.

Bisma Farooq has been a freelance journalist in India for the past five years. She turned to Artificial Intelligence training and data annotation for Invisible Technologies to supplement income, but found the commission-based work technical, repetitive and creatively draining. Bettina Blass has been a freelance journalist in Germany since 2003. For a year and a half, Blass worked with the startup on an Artificial Intelligence editing tool that automatically turns incoming press releases, police alerts or community notices into journalistic news articles, writing prompts that taught the system basic news structures such as the inverted pyramid.

The pay picture varied sharply. Khan and Osborne found tech-company work more financially stable and more lucrative than freelancing, while Blass and Farooq said their Artificial Intelligence assignments paid less than journalism. The ethical tension was also clear: several journalists worried they were helping build systems that could compete with them, even as the work gave them experience in a changing labor market.

Close exposure to the technology left the journalists skeptical that machines can fully replace human reporting. They described Artificial Intelligence systems as dependent on human labor, prone to bias, errors and hallucinations, and unable to reproduce the judgment, creativity and emotional perspective central to journalism. The most optimistic view cast Artificial Intelligence as useful for routine tasks such as press releases or sports scores, freeing reporters to focus on higher-value work while requiring closer collaboration between engineering and editorial teams.

The economics are presented as especially fragile. In reality, 89%+ of all Artificial Intelligence revenues and 90%+ of all compute demand comes from two companies, OpenAI and Anthropic, largely based on money-losing subsidized Artificial Intelligence subscriptions and unrestrained token burn. Enterprises are already capping their Artificial Intelligence spend after multiple companies blew through their annual token budgets in a few months. The total, actual revenue of the entire Artificial Intelligence industry, including OpenAI, Google, Microsoft, Amazon, and Anthropic, has barely reached $100bn in 2026. That includes every ounce of compute spend, every penny of the $500mn that a single customer accidentally spent on Anthropic’s API, and every cent of NVIDIA’s backstop deal with CoreWeave.

Technical limits are tied to hallucinations, with generative Artificial Intelligence described as probabilistic software that guesses at outputs rather than knowing facts or making reasoned decisions. Coding is treated as the strongest claimed use case, yet the benefits are portrayed as unclear: not reliably saving money, not clearly saving time, and not consistently improving products. Outside coding, examples are described as vague, including possible agents at Goldman Sachs and Novo Nordisk integrating ChatGPT models to analyze complex data sets.

The infrastructure buildout is framed as another sign of financial overreach. Major data center borrowing included $178.5bn in data center debt deals in the US in 2025 and $50bn in data center construction in April 2026 alone. Data centers take anywhere from 18 to 36 months to build, with Microsoft finishing a grand total of zero of the data centers it broke ground on in 2023, and JP Morgan saying a month ago that 60% of capacity planned for completion in 2027 hasn’t even started construction, with another 7% delayed. Despite the supposed 100GW+ of data center capacity being planned, Artificial Intelligence compute demand is described as concentrated around Anthropic and OpenAI. NVIDIA’s continued growth relies on a dwindling subset of clients, with 54% of its last quarter’s revenue and 64% of its accounts receivable coming from three customers in its last quarterly earnings.

Under the decision, Google will have to give publishers “effective tools” to prevent their content from being used to power the company’s generative Artificial Intelligence services and its search features, including Artificial Intelligence Overviews and Artificial Intelligence Mode. Google will also have to properly cite publisher content in Artificial Intelligence-generated search results by using clear links. Publishers must also be allowed to opt out of having their content used to fine-tune Artificial Intelligence models.

The watchdog said the decision will give publishers a stronger hand when negotiating content deals with Google. Publishers are defined as anyone who puts content on the web that is available to people in Britain. The ruling was expected because the regulator had released draft proposals at the start of the year after using its new digital powers to label Google a “strategic” player in online search advertising.

The Competition and Markets Authority previously found that news publishers had suffered a drop in traffic since Google rolled out Artificial Intelligence Overviews, summaries that appear at the top of some search queries, because fewer users are clicking through to the original articles. The requirements will also apply to major changes Google unveiled in May that further embed Artificial Intelligence in the company’s search services. Google said it is engaging with regulators to ensure website owners have the right tools as user preferences evolve, and said it is testing a new control for how links and content appear in generative Artificial Intelligence search features.