Development

Gemini Nano boosts scam detection on Chrome

Google has released a new report outlining how it is using AI to better protect users from online scams across its platforms.

The company says AI is now actively fighting scams in Chrome, Search and Android, with new tools able to detect and neutralise threats more effectively than before.

At the heart of these efforts is Gemini Nano, Google’s on-device AI model, which has been integrated into Chrome to help identify phishing and fraudulent websites.

The report claims the upgraded systems can now detect 20 times more harmful websites, many of which aim to deceive users by creating a false sense of urgency or offering fake promotions. These scams often involve phishing, cryptocurrency fraud, clone websites and misleading subscriptions.

Search has also seen major improvements. Google’s AI-powered classifiers are now better at spotting scam-related content before users encounter it. For example, the company says it has reduced scams involving fake airline customer service agents by over 80 per cent, thanks to its enhanced detection tools.

Meanwhile, Android users are beginning to see stronger safeguards as well. Chrome on Android now warns users about suspicious website notifications, offering the choice to unsubscribe or review them safely.

Google has confirmed plans to extend these protections even further in the coming months, aiming to cover a broader range of online threats.

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

OpenAI launches data residency in India for ChatGPT enterprise

OpenAI has announced that enterprise and educational customers in India using ChatGPT can now store their data locally instead of relying on servers abroad.

The move, aimed at complying with India’s upcoming data localisation rules under the Digital Personal Data Protection Act, allows conversations, uploads, and prompts to remain within the country. Similar options are now available in Japan, Singapore, and South Korea.

Data stored under this new residency option will be encrypted and kept secure, according to the company. OpenAI clarified it will not use this data for training its models unless customers choose to share it.

The change may also influence a copyright infringement case against OpenAI in India, where the jurisdiction was previously questioned due to foreign server locations.

Alongside this update, OpenAI has unveiled a broader international initiative, called OpenAI for Countries, as part of the US-led $500 billion Stargate project.

The plan involves building AI infrastructure in partner countries instead of centralising development, allowing nations to create localised versions of ChatGPT tailored to their languages and services.

OpenAI says the goal is to help democracies develop AI on their own terms instead of adopting centralised, authoritarian systems.

The company and the US government will co-invest in local data centres and AI models to strengthen economic growth and digital sovereignty across the globe.

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

CrowdStrike cuts jobs amid AI shift

Cybersecurity firm CrowdStrike is laying off 500 employees—5% of its workforce—as it shifts towards an AI-led operating model to boost efficiency and hit a $10 billion annual revenue goal.

In a letter to staff, CEO George Kurtz described AI as a ‘force multiplier’ meant to reduce hiring needs instead of expanding headcount.

The restructure, expected to cost up to $53 million through mid-2026, will still see hiring in customer-facing and engineering roles.

Yet despite its optimism, the company’s regulatory filings flag notable risks in depending on AI, such as faulty outputs, legal uncertainty, and the challenge of managing fast-moving systems. Analysts have also linked the shift to wider market pressures, not merely strategic innovation.

Principal analyst Sofia Ali warned that the AI-first approach may backfire if transparency, governance, and human oversight are not prioritised. Over-reliance on automation—especially in threat detection or customer support—could erode user trust instead of reinforcing it, particularly during critical incidents.

CrowdStrike’s move mirrors a broader tech trend: over 52,000 tech jobs were cut in early 2025 as firms embraced AI to replace automatable roles. For cybersecurity leaders, the challenge now lies in balancing AI’s promise with the human expertise essential to trust and resilience.

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

Google boosts AI coding and video skills with Gemini 2.5 Pro

Google has unveiled Gemini 2.5 Pro Preview (I/O edition), its latest AI model update, ahead of the annual I/O developer conference.

The new version introduces upgrades designed to enhance coding assistance, video comprehension, and web development capabilities, further highlighting the company’s commitment to integrating generative AI across its products.

According to Google, Gemini 2.5 Pro delivers noticeable improvements in software development workflows. The AI model now ranks first on the WebArena leaderboard, which assesses the quality of AI-generated websites through human feedback.

Google reports a 147-point increase in Elo score compared to the previous version. Enhanced features also include better code interpretation and generation, real-time function calling, and lower rates of silent failures in tasks such as bug fixing and editing.

Video analysis is another key area of progress. Gemini 2.5 Pro achieved an 84.8% score on the VideoMMe benchmark, which evaluates AI performance in understanding and answering questions about visual content, particularly educational videos.

This marks another step in Google’s multimodal AI ambitions, which seek to unify text, images, code, audio, and video into cohesive AI systems. However, Google has not detailed how privacy and content safety are managed when processing video data.

Developers can now access Gemini 2.5 Pro via the Gemini API, Google AI Studio, and Vertex AI, as well as through Gemini’s mobile and web apps.

Currently released as a preview, the model will likely receive updates following developer feedback during and after the I/O event. Details about general availability or potential pricing changes have not been disclosed.

The release of Gemini 2.5 Pro comes as AI competition intensifies, with OpenAI, Anthropic, and Meta also advancing their foundation models. While Google highlights Gemini’s strong performance in coding and web development, broader comparisons between models remain limited for now.

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

Chennai team wins ₹50 lakh at Agentforce Hackathon for AI hotel solution

AI took centre stage at the Agentforce Hackathon 2025 during TrailblazerDX in Bengaluru, where a Chennai-based team from Bounteous x Accolite Salesforce claimed the grand prize of ₹50 lakhs. Their AI-powered project impressed judges and attendees, standing out among 195 competing teams.

The team, calling themselves ‘Aichemist’, included Mayur Kinhekar, Rajaseeman PS, Yogeshwar Andi Sudhakaran, Sheetalraj Gangadhar, and Vijay Kalidasan. Together, they set out to revolutionise hospitality using Salesforce’s cloud platform.

Their Smart Hotel Agent uses data cloud, loyalty tiers, Slack, and service tablets to deliver seamless, personalised hotel experiences. ‘We are thrilled with this achievement,’ said senior consultant Mayur Kinhekar. ‘We are proud to be AI agents helping shape the future of AI.’

Team lead Rajaseeman PS explained that their goal was to reduce the need for human interaction in routine hotel tasks. By giving guests tablets, they created a solution that manages everything from check-in to check-out with ease.

In one example, Rajaseeman noted how a spill in a hotel room could be instantly reported and resolved through automated notifications sent to the appropriate department, bypassing traditional calls to the front desk.

By using AI to streamline services, the team not only improved guest experiences but also freed up hotel staff to focus on critical tasks. Their innovative approach captured the imagination of the hackathon’s judges and earned them the prestigious top spot and grand prize.

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

AWS expands in Latin America with first Chile cloud region

Amazon Web Services (AWS) will invest $4 billion to build its first cloud region in Chile by late 2026, marking a significant expansion of its Latin American presence.

The new region will feature three availability zones and become AWS’s third region in the region after Brazil and Mexico, and its 37th worldwide.

The company confirmed that all necessary permits for construction and operation have been secured. AWS expects the Chilean region to provide substantial computing power for generative AI, data analytics and enterprise applications.

The decision reflects the growing demand for low-latency cloud services as Chile’s cloud market is projected to expand more than 30 percent in 2024 and reach $1.9 billion by 2025. IDC forecasts continued growth at about 20 percent annually through 2028.

Local organisations, including LATAM Airlines, AgroSuper and Andrés Bello University, already rely on AWS for critical workloads. Partners such as Deloitte, Accenture and NTT will help support customer onboarding and manage systems.

AWS’s expansion follows its entry into Chile in 2019 with a content delivery edge location, followed by the addition of Outposts, Direct Connect, and a Local Zone over the past few years.

Environmental considerations remain central to the project. AWS will limit water cooling to only 4 percent of the year — equivalent to the annual consumption of two average Chilean households — and primarily use air and evaporative cooling.

The company reached 100 percent renewable-energy usage in 2023 and targets net-zero carbon emissions by 2040 as part of its Climate Pledge.

Competition is heating up in Chile’s cloud market. Microsoft Azure plans to open its local region this year, while Google faced regulatory setbacks after a court partially revoked its permit for a $200 million data centre project.

AWS hopes its early investment will help it capture a larger share of the Latin American cloud services sector, reinforcing its global network of 36 regions and 114 availability zones.

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

Apple may replace Google with AI in Safari

Apple may soon reshape how users search the web on iPhones and other devices by integrating AI-powered search engines directly into Safari instead of relying solely on Google.

According to Bloomberg, the company is ‘actively looking at’ expanding options in its browser to include AI systems such as OpenAI’s ChatGPT and Perplexity, potentially disrupting Google’s long-held dominance in online search.

Currently, Google pays Apple around $20 billion a year to remain the default search engine in Safari — about 36% of the search ad revenue generated through Apple devices. But that relationship may be under pressure, especially as AI tools gain popularity.

Apple has already partnered with OpenAI to bring ChatGPT into Siri, while Google is now pushing to include its Gemini AI system in future Apple products.

Alphabet’s shares dropped 6% following the news, while Apple saw a 2% dip. Apple executive Eddy Cue, testifying in an ongoing antitrust case, noted a recent decline in Safari searches and said he expects AI search tools to eventually replace traditional engines like Google.

Apple, he added, plans to introduce these AI services as built-in alternatives in Safari in the near future.

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

LockBit ransomware Bitcoin addresses exposed

Nearly 60,000 Bitcoin addresses linked to LockBit’s ransomware operations have been exposed following a major breach of the group’s dark web affiliate panel.

The leak, which included a MySQL database dump, was shared publicly online and could assist blockchain analysts in tracing LockBit’s financial activity instead of leaving such transactions untracked.

Despite the scale of the breach, no private keys were leaked. A LockBit representative reportedly confirmed the incident in a message, stating that no sensitive access data was compromised.

However, the exposed database included 20 tables, such as one labelled ‘builds’ that contained details about ransomware created by affiliates and their targeted companies.

Another table, ‘chats,’ revealed over 4,400 messages from negotiations between victims and LockBit operators, offering a rare glimpse into the inner workings of ransomware extortion tactics.

Analysts believe the hack may be connected to a separate breach of the Everest ransomware site, as both featured identical messages, hinting at a possible link.

The incident has again underscored the central role of cryptocurrency in the ransomware economy. Each victim is typically given a unique address for payments, making tracking difficult.

Instead of remaining hidden, these addresses now give law enforcement and blockchain experts a chance to trace payments and potentially link them to previously unidentified actors.

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

Netflix introduces AI chatbot to help you pick what to watch

Netflix is trialling an AI chatbot inside its iOS app, offering a new way for users to find content by simply typing natural phrases instead of relying on standard searches. In this small, opt-in beta, users might say things like ‘I want something funny and upbeat; to receive tailored recommendations.

The company believes the AI chatbot could soon become a core part of its app on both iOS and Android, and perhaps even land on TVs in future.

Alongside this, Netflix is reshaping the user experience by surfacing helpful labels like ‘Emmy Award Winner’ and ‘#1 in TV Shows’ to help viewers choose faster instead of scrolling endlessly.

Search and My List are moving to the top of TV screens for better visibility, and the homepage is getting a cleaner, more modern design.

Netflix says recommendations will also shift dynamically based on a viewer’s mood or interests, although it hasn’t explained exactly how this will work.

On mobile, Netflix plans to roll out a vertical feed of show and movie clips in the coming weeks. You’ll be able to tap to watch, save, or share immediately—turning content discovery into a quick and interactive experience instead of a chore.

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

Mistral AI unveils enterprise tools to rival Big Tech

French startup Mistral AI is making waves with the launch of its Medium 3 model family and a business chatbot called Le Chat Enterprise, aiming to disrupt a market long dominated by OpenAI, Google, Microsoft, and Anthropic.

The company claims its new models outperform rivals while needing fewer computational resources, giving businesses a more efficient and affordable AI alternative.

The Medium 3 family consists of Small, Standard and Large variants, with the Standard version described as the ideal balance between power and efficiency.

Capable of handling context windows up to 128,000 tokens, these models can process long documents and complex inputs without the costs typically associated with top-tier AI systems.

Meanwhile, Le Chat Enterprise offers companies a privacy-focused, ready-to-deploy chatbot tailored for business use instead of consumer settings.

Backed by €385 million in funding and a team of former DeepMind and Meta researchers, Mistral continues to set itself apart by combining open-source releases with commercial offerings.

If its performance claims hold in independent benchmarks, Mistral may force the AI industry to compete not only on raw capability but also on value—a shift that could benefit enterprises seeking practical, cost-effective AI solutions instead of hype-driven tools.

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

Diplo chatbot can make mistakes. Consider checking important information.