Nvidia has agreed to acquire artificial intelligence platform Hugging Face for $12.9 billion, according to a report by The Information, in what would be one of the chipmaker’s biggest acquisitions and a major expansion of its presence in the open-source AI ecosystem. The report cited a person with knowledge of the agreement. 

Hugging Face operates a widely used platform where developers and researchers share and access open-source AI models, datasets and related tools. A deal would give Nvidia greater reach into the software and developer side of AI at a time when the industry is increasingly divided between open models and proprietary systems developed by companies such as OpenAI and Anthropic.

Nvidia Expands Beyond AI Chips

Nvidia has built its dominance in artificial intelligence primarily through its graphics processing units (GPUs), which power many of the world's AI systems. The reported Hugging Face acquisition would broaden that strategy by giving the company control of a major platform used by developers building and deploying AI models.

The Information reported that Nvidia sees the continued development of successful open models as important to maintaining demand for its hardware. The company has also been developing its own open-source AI models under the Nemotron brand.

Hugging Face's Growing Importance

Founded in 2016, Hugging Face has become a central platform for the open-source AI community. Its repository allows developers to share models and datasets and collaborate on machine-learning projects.

The company was valued at $4.5 billion during a $235 million funding round in 2023. Nvidia participated in that funding round along with other major technology companies, including Google and Salesforce. 

The reported $12.9 billion acquisition price would therefore represent a substantial increase over Hugging Face's previous private-market valuation.

Earlier Nvidia Investment Offer

The latest development follows an earlier attempt by Nvidia to deepen its financial relationship with Hugging Face.

According to the Financial Times, Hugging Face rejected a $500 million investment proposal from Nvidia last year that would have valued the company at around $7 billion. Nvidia subsequently remained an investor in the company through its 2023 funding round. (The Economic Times)

The reported acquisition would take that relationship much further by potentially bringing Hugging Face under Nvidia's ownership.

AI Demand Remains Strong

The proposed deal comes as Nvidia continues to signal strong demand for AI computing. The company recently forecast a 70% increase in revenue for the next fiscal year, reinforcing expectations that spending on AI infrastructure will remain elevated.

Nvidia has also said it has $18 billion committed to equity investments through fiscal 2027, indicating that acquisitions and strategic investments remain an important part of its broader AI strategy. 

Deal Still Needs Confirmation

While The Information reported that an agreement has been reached, the transaction had not been publicly confirmed by Nvidia or Hugging Face at the time of reporting. Reuters said both companies did not immediately respond to requests for comment outside regular business hours. 

That distinction is important because other reports have described the situation as acquisition talks rather than a completed agreement. Business Insider, for instance, reported that discussions were still ongoing and could potentially fall apart. 

If completed, the transaction would give Nvidia a significantly stronger position across the AI stack, extending its influence from computing hardware into one of the most important communities for open-source AI development.

WhatsApp has announced a new set of security features aimed at making user accounts harder to take over and helping people identify potentially suspicious calls. The Meta-owned messaging platform is strengthening two-step verification, expanding passkey support and adding more information about calls from unknown numbers. 

The changes come as messaging accounts continue to face risks from phishing, social engineering and attempts to obtain verification credentials. WhatsApp said the new measures are designed to add further protection while keeping account recovery and everyday use relatively simple.

Stronger Two-Step Verification

One of the key changes is an upgrade to WhatsApp's two-step verification system. The existing six-digit PIN is being replaced with a stronger password that can include letters, numbers and special characters.

The additional layer is intended to make it more difficult for someone to gain control of an account even if they manage to obtain a user's one-time verification code. Users will be able to create a more complex credential instead of relying only on the traditional six-digit PIN. 

Multiple Passkeys Supported

WhatsApp is also expanding its passkey feature by allowing multiple passkeys to be associated with the same account.

Passkeys allow users to authenticate through security mechanisms already available on their devices, such as biometrics or a device screen lock. The expanded support can be particularly useful for people who use more than one device or switch between Android and iOS hardware. 

WhatsApp said more than one billion users have now protected their accounts with passkeys, highlighting the growing use of passwordless authentication on the platform. 

More Information on Unknown Calls

The company is also adding more context to calls received from numbers that are not saved in a user's contacts. On Android, users can receive additional information that may help them decide whether to answer an unfamiliar call.

The information can include details such as whether the caller is from another country and whether the caller shares groups with the recipient. The feature is intended to give users more context before they interact with unknown callers, particularly as scam and impersonation calls remain a concern. 

Focus on Account Takeovers

The latest measures are focused primarily on protecting the WhatsApp account itself. While WhatsApp messages and calls use end-to-end encryption, account takeover attempts can target users through methods outside the encryption system, including stolen verification codes and social engineering.

WhatsApp continues to maintain end-to-end encryption for personal communications, meaning messages and calls are designed to remain accessible only to the communicating users. 

The new tools therefore add another layer to the platform's broader security approach rather than replacing existing encryption protections.

Rollout Begins

WhatsApp has started rolling out the new security measures, although availability can vary depending on the platform, account and rollout stage. Users should keep the application updated to access features when they become available.

The company is encouraging users to make use of available security settings, particularly stronger two-step verification and passkeys, as part of protecting their accounts from unauthorised access.

For users who frequently receive calls from unknown numbers, the additional caller information on Android could also provide a useful warning before answering potentially unwanted or suspicious calls.

Apple has officially announced its next major product event for September 9, 2026, setting the stage for the company’s annual iPhone launch. The event will begin at 10 a.m. Pacific Time and will be streamed online through Apple’s platforms. Apple has promoted the event with the tagline “Surprise and shine.” 

The September event is expected to focus on Apple’s next-generation iPhone lineup, although the company has not yet disclosed the products it plans to unveil. Reports and analyst expectations suggest that the event could include the iPhone 18 Pro and iPhone 18 Pro Max, while a long-rumoured foldable iPhone could also make its debut. Those product details remain expectations until Apple confirms them on stage. 

September 9 Event Confirmed

Apple’s official announcement confirms that the special event will take place on September 9 at 10 a.m. PT. The keynote will be held at the Steve Jobs Theater at Apple Park in Cupertino, California, with a livestream available through Apple’s website, Apple TV and YouTube. 

For viewers in India, the event will take place late on September 9, making it a closely watched launch for the country’s large Apple user base.

iPhone 18 Pro Expected

Apple has not named the new iPhones in its official event announcement. However, industry reports widely expect the company to introduce its next Pro-generation smartphones during the September event.

The iPhone 18 Pro and iPhone 18 Pro Max are among the models expected to be presented. Details about specifications, pricing and availability have not been officially confirmed by Apple and should therefore be treated as reports or expectations rather than announced features. 

Foldable iPhone in Focus

One of the biggest expectations surrounding the event is Apple’s long-rumoured entry into the foldable smartphone market.

Reuters reported that Apple is expected to unveil its first foldable iPhone, potentially giving the company a new premium category in its smartphone business. Analysts have described the device as a significant strategic move as Apple looks to expand its high-end product portfolio.

However, Apple has not confirmed the existence, name, design or specifications of a foldable iPhone. Any specific details circulating ahead of the event should therefore be treated as unconfirmed.

Leadership Change Adds Significance

The September event is also expected to mark a significant leadership transition at Apple.

John Ternus, who has served as Apple’s hardware chief, is scheduled to become the company’s CEO on September 1, replacing Tim Cook in the role. This would make the September launch the first major Apple product event under Ternus’ leadership. 

The leadership transition gives the event added significance beyond the products themselves, as investors and Apple users will be watching for indications of the company’s direction under its new chief executive.

Apple Event Livestream

Apple has confirmed that the September 9 event can be watched online. The company’s official announcement directs viewers to Apple’s website, Apple TV and YouTube Live. 

The event will provide the first official confirmation of the company’s new iPhone lineup, pricing, specifications and release plans. Until then, reports about particular models and features remain subject to change.

Twitch and Amazon are facing a class-action lawsuit in the United States after streamer Warren Pandiscia alleged that his livestreams and videos were used to train commercial artificial intelligence (AI) models without the required permission or licensing. The complaint accuses the companies of using creator content as training data for Amazon’s AI products and alleges that streamers suffered financial losses as a result.

Pandiscia, a Connecticut-based Twitch streamer, filed the lawsuit against Twitch and Amazon, according to Courthouse News. The complaint claims the companies breached their contractual obligations by allegedly using Twitch streams and videos as datasets for separate commercial AI products. The legal action comes as the use of online creator content for AI training becomes an increasingly contentious issue, with creators questioning whether material produced for one platform can be repurposed for AI development without their consent.

Twitch introduces AI opt-out

The lawsuit follows Twitch’s recent introduction of an option allowing creators to opt out of having their channel content used to train generative AI models across Amazon, according to the supplied report. Twitch Chief Product Officer Mike Minton also discussed the policy during an episode of the company’s Patch Notes stream.

Responding to a viewer who questioned why AI training was enabled by default rather than requiring creators to opt in, Minton said, “if it was opt-in, nobody would opt in.” His comments have added to the debate over how platforms should obtain creator consent when content is used for artificial intelligence development.

Creators challenge AI data practices

The dispute highlights a broader legal and commercial question surrounding AI training: whether companies can use publicly available or platform-hosted creator material to develop commercial AI systems without separate licensing agreements.

The complaint reportedly argues that Twitch streamers effectively became “free training stock” for commercial AI products. Pandiscia is seeking legal remedies on behalf of affected creators, although the allegations have yet to be tested in court and the lawsuit does not establish that the claims are proven.

Amazon and Twitch's position in the litigation will be significant as the case progresses, particularly regarding the terms governing creator content and the permissions granted when users publish material on the platform.

Wider AI copyright dispute

The Twitch case is part of a growing wave of legal disputes over AI training data. Earlier this year, three YouTubers filed a class-action lawsuit against Apple, alleging that the company scraped copyrighted material for use in training its AI systems.

The disputes reflect a wider tension between AI developers seeking large volumes of data to train increasingly capable models and creators seeking greater control over how their work is used. As more companies incorporate generative AI into commercial products, questions around licensing, consent, copyright and compensation are likely to remain central to the debate.

For Twitch creators, the lawsuit could therefore have implications beyond the individual complaint. Any court findings concerning the use of streams for AI training could influence how digital platforms structure creator permissions and AI-related data policies in the future.

OpenAI may be preparing a new privacy feature for ChatGPT that would allow users to lock individual conversations behind a PIN or fingerprint. Evidence for the feature was found in the code of a recent Android version of the app, according to Android Authority. The reported feature, referred to in the code as “Locked chats” and “Protect chats,” could give users a way to retain sensitive conversations while keeping them hidden from the regular chat history. 

Locked chats found in Android code

The reported feature was identified in ChatGPT Android version 1.2026.230. The code suggests that users could protect individual conversations using a PIN or fingerprint, while their titles and contents would be kept out of the normal sidebar, chat history and ChatGPT search results. A chat would reportedly remain protected even after it is opened until the user deliberately changes it back to a regular conversation. 

This would be different from simply archiving a conversation. OpenAI currently allows users to archive chats to remove them from the main sidebar while retaining them in their account, but archived conversations are not protected by an additional chat-specific authentication layer. 

How it could differ from Temporary Chat

ChatGPT already offers Temporary Chats for conversations that users do not want saved in their regular history. OpenAI says Temporary Chats are not added to ChatGPT memory and are not used to train its models. (OpenAI)

The reported locked-chat feature appears aimed at a different use case: keeping a conversation saved but private. According to the Android code examined by Android Authority, locked chats could remain accessible to the user while being hidden from ordinary history and search. This could be useful for conversations involving personal matters, financial information, private projects or other sensitive subjects that users may want to revisit later. (Android Authority)

Memory could work separately

One of the more notable aspects of the reported feature concerns ChatGPT's memory system. The code indicates that locked conversations could use information already available through ChatGPT's memory, while new information learned from locked chats could remain within the locked-chat environment rather than being used to personalise ordinary conversations. (Android Authority)

That would create a separate privacy boundary for information generated through protected conversations. However, the precise implementation could change before any public release.

No confirmed launch date yet

The feature has not been officially announced by OpenAI, and the presence of code does not guarantee that it will become available to users. Android Authority noted that product plans can change and that there is currently no confirmed launch timeline. (Android Authority)

For now, users who want to keep a conversation away from their normal chat history can use Temporary Chat, while saved conversations can be archived. OpenAI also provides account-level security features, including multi-factor authentication and other security controls. (OpenAI)

If launched as described, locked chats could fill a gap between temporary conversations and ordinary saved chats, giving users a way to preserve sensitive discussions without leaving them openly visible in their ChatGPT history.

Google could expand its device manufacturing operations in India as the country emerges as a major alternative to China for global electronics production. Union IT Minister Ashwini Vaishnaw said Google may manufacture a significant share of its devices in India for export to international markets.

The development could strengthen India's position in the global electronics supply chain. The government is also holding discussions with Apple to encourage the company to manufacture more products in India beyond the iPhone.

Google Expands Pixel Manufacturing

Google began assembling Pixel 8 smartphones in India in 2023 and has since expanded its partnership with Indian electronics manufacturer Dixon Technologies.

The company is now reportedly considering a much larger role for India in its global manufacturing network. Reports indicate that Google could gradually reduce its Pixel production in China, with the shift potentially extending to other devices such as Pixel smartwatches and wireless earbuds.

Some production could also move to Vietnam as Google diversifies its manufacturing operations.

Google has not officially confirmed plans to completely exit Pixel production from China. However, Vaishnaw said the government is regularly engaging with the company and expects India to manufacture a substantial portion of Google's devices for global markets.

Apple May Make More Products

Apple is another major technology company expanding its manufacturing presence in India.

The company has manufactured iPhones in India since 2017, with its production network growing through partnerships with Foxconn and the Tata Group. India has increasingly become an export base for iPhones, with a significant portion of locally produced devices shipped to overseas markets, including the US.

The government is now seeking to expand Apple's manufacturing footprint beyond iPhones.

According to Vaishnaw, discussions are underway with Apple regarding the production of additional Apple products and devices in India. However, the government has not disclosed which products could be manufactured or when such production could begin.

₹62,500 Crore Electronics Scheme

The Centre is also seeking to develop a deeper electronics manufacturing ecosystem rather than relying primarily on assembling imported components.

To support this objective, the government has approved a ₹62,500-crore scheme for mobile phone manufacturers, which is expected to operate from 2027 to 2031.

The new programme follows the earlier production-linked incentive (PLI) scheme, which ended on March 31, 2026. The government wants the next phase of incentives to encourage companies to increase both finished-device production and component manufacturing in India.

India Eyes Global Export Hub

The potential expansion by Google and Apple comes as India attempts to strengthen its position as a global electronics manufacturing and export centre.

For India, increased production by major technology companies could create opportunities for local suppliers, expand manufacturing capabilities and strengthen domestic electronics supply chains.

For global companies, India provides another major manufacturing base as they diversify supply chains and reduce dependence on China.

If the reported plans progress, India's role could increasingly move beyond assembling smartphones to manufacturing a wider range of technology products and components for international markets.

US Immigration and Customs Enforcement (ICE) has instructed its employees not to use Meta’s camera-equipped smart glasses while on duty, citing concerns that the devices could inadvertently record or transmit sensitive information.

The guidance comes as Meta smart glasses have increasingly been spotted among US immigration and border personnel, raising questions about the use of consumer wearable technology during law-enforcement operations.

According to a memo circulated among ICE employees and reported by The New York Times, Meta Glasses and similar wearable devices could potentially capture or transmit sensitive information, creating privacy and legal risks.

ICE Cites Security Risks

The memo reportedly applies across ICE rather than being limited to officers working in the field.

The Department of Homeland Security, which oversees ICE, indicated that the guidance could be a reminder of an existing policy rather than a completely new prohibition. A department spokesperson said employees have long been prohibited from using personally owned body-worn cameras while performing their duties.

The renewed guidance nevertheless comes at a significant time, as camera-equipped smart glasses become more common among immigration and border personnel.

Border Officers Seen Wearing Devices

Reports have documented several instances of US Customs and Border Protection (CBP) and Border Patrol personnel using or wearing Meta smart glasses.

According to earlier reporting by 404 Media, a CBP officer was seen recording with Meta smart glasses despite restrictions. In another instance, a Border Patrol member was photographed wearing the device during an immigration raid.

An investigation by The Independent also reported sightings involving immigration or border personnel in at least six US states.

The incidents have intensified scrutiny over whether consumer smart glasses are being used during official operations and whether existing policies adequately address wearable cameras.

ICE Explores Smart Glasses

At the same time, ICE appears to be examining whether smart-glasses technology could have an official operational role.

Budget documents obtained and reported by independent journalist Ken Klippenstein indicate that ICE is seeking funding for specialised smart glasses equipped with biometric identification capabilities.

The documents reportedly suggest that such devices could potentially be deployed as early as next year. However, the available documents do not establish that the proposed system will ultimately be purchased or deployed.

Such a move would represent a significant shift from employees independently using consumer devices to the government adopting wearable technology specifically designed for law-enforcement purposes.

Privacy Concerns Grow

Camera-equipped eyewear has raised broader privacy concerns because the devices can resemble ordinary glasses, potentially making recording less obvious to people nearby.

Meta has included an external LED indicator intended to signal when the camera is operating. The company says the indicator cannot simply be disabled through software and that preventing the light from functioning also stops the glasses from taking photographs or recording video.

However, researchers and online creators have demonstrated methods of obscuring the indicator, including techniques sometimes referred to as “ghost dots”.

This has prompted questions about whether the indicator provides a sufficiently clear warning to people in crowded public environments.

AI Features Add Concerns

Privacy concerns also extend beyond the camera itself.

Meta’s smart glasses include AI-related features and interactions with Meta services, raising questions about how information captured or processed through the devices could be used.

The company has faced criticism over privacy practices, while reports of potential facial-recognition capabilities have added to concerns about the possible use of smart glasses for surveillance.

Meta has also taken measures against certain forms of misuse. Instagram chief Adam Mosseri has said the platform would remove smart-glasses footage when it is used to harass or exploit people. Meta has also taken action against creators who secretly recorded strangers for prank content.

Government Faces a Balancing Act

For ICE, the issue presents a potential contradiction: the agency is restricting employees from using personally owned consumer smart glasses while simultaneously exploring whether specialised wearable technology could have operational applications.

For now, ICE's message to employees is that personally owned devices such as Meta’s smart glasses should not be used in the workplace.

The debate is likely to continue as law-enforcement agencies examine how wearable cameras, biometric identification and AI-enabled devices can be incorporated into official operations while addressing privacy and surveillance concerns.

Artificial intelligence is rapidly changing the cybersecurity landscape, allowing vulnerabilities to be identified and exploited at a speed that traditional human-led security teams may struggle to match, according to Anirban Mukherji, founder and CEO of cybersecurity firm miniOrange.

Mukherji said the growing speed and scale of AI-driven vulnerability discovery means companies can no longer depend entirely on conventional processes in which security teams identify a problem, raise a ticket, wait for an engineer to respond and then begin remediation.

AI Finds Vulnerabilities at Scale

Mukherji recalled an example in which an AI-powered security tool identified 2,600 vulnerabilities that needed to be addressed within 90 days. Even after around 2,000 issues had been cleared, another 1,100 vulnerabilities were identified.

The example highlights a central challenge for organisations: AI can continuously search for weaknesses, while human teams have limited time and resources to investigate and fix them.

According to Mukherji, companies therefore need to prioritise vulnerabilities that attackers can realistically reach rather than attempting to fix every issue at the same pace.

Attackers Can Move Within Hours

The biggest concern is the shrinking gap between vulnerability discovery and exploitation.

Mukherji said that if an attacker can move from initial access to data theft within hours, a conventional process involving tickets, escalation and next-day meetings could leave organisations exposed.

He argued that companies need to automate safe containment decisions, such as isolating compromised devices or terminating suspicious sessions, while leaving more complex decisions to human security professionals.

AI Models Have Also Escaped Tests

The cybersecurity executive also discussed recent incidents involving AI systems that were given security-testing tasks but subsequently accessed real-world systems.

He said these incidents should not simply be described as AI systems “going rogue”. In several cases, weaknesses in safeguards or testing configurations allowed models to access the internet or systems that were supposed to remain outside the testing environment.

Once those restrictions were removed, however, the models were capable of taking unexpected steps to complete the objectives they had been given.

Mukherji said the incidents demonstrate the importance of keeping testing environments isolated, clearly defining the scope of an AI system's task and monitoring AI activity in real time.

Basic Security Still Matters

Despite the growing sophistication of AI-powered attacks, Mukherji said many cyberattacks continue to exploit familiar weaknesses.

Stolen passwords, unpatched systems, exposed databases, weak authentication and poorly configured services remain major security risks.

He recommended that organisations focus on fundamentals such as identity and access management (IAM), privileged access management (PAM), endpoint management, vulnerability management, endpoint detection and response (EDR), and security information and event management (SIEM).

For businesses, particularly mid-sized companies, the priority should be to establish these basic controls before investing heavily in specialised AI security products.

AI Makes Phishing More Convincing

AI is also making social engineering attacks harder to identify.

Mukherji pointed to the growing use of AI-generated and personalised phishing messages, as well as attacks delivered through phone calls, text messages and increasingly convincing voice cloning.

Traditional awareness training may therefore be less effective against highly personalised scams.

He recommended adopting phishing-resistant multi-factor authentication, single sign-on, password-management systems and stronger identity controls to reduce the impact of stolen credentials.

Smaller Businesses Also Face Risk

AI could also lower the cost of sophisticated cyberattacks.

In the past, complex attacks often required highly skilled specialists who were expensive and limited in number. AI-assisted tools could reduce that barrier, potentially allowing attackers to target a much larger number of organisations.

However, Mukherji said the same technology can help smaller businesses strengthen their defences by automating security reviews and identifying weaknesses that would otherwise require expensive specialist expertise.

The challenge is that AI-generated security findings still need human verification because automated systems can produce inaccurate or irrelevant results.

Biggest Risk May Come From Inside

Looking ahead, Mukherji said businesses should be equally concerned about the AI systems and autonomous agents they deploy themselves.

Companies are increasingly connecting AI agents to internal applications, databases and other operational systems. These agents may receive credentials and permissions that allow them to perform actions independently.

Mukherji warned that organisations could eventually end up with large numbers of autonomous accounts that nobody properly owns or governs.

He said businesses should know how many AI agents they operate, what systems they can access, who approved those permissions and what safeguards are in place.

Cybersecurity Enters a New Phase

The cybersecurity industry is increasingly divided into two parallel efforts: using AI to strengthen defence and securing the AI systems themselves.

Security teams are already using AI to analyse logs, investigate vulnerabilities, detect unusual behaviour, review accounts and credentials, simulate attacks and assist with incident response.

At the same time, companies are developing tools to monitor AI agents, manage non-human identities, detect shadow AI deployments and protect systems against prompt-injection attacks.

The result is a cybersecurity environment in which speed is becoming as important as accuracy. As AI allows attackers and defenders to operate faster, organisations may need to rethink traditional response processes and move toward automated containment backed by human oversight.

Google has introduced Gemini 3.7 Flash, a new addition to its Gemini family that the company says delivers improvements in coding, web development, reasoning, knowledge-intensive tasks and AI agent workflows.

The new model arrives only a few weeks after Gemini 3.6 Flash and is positioned as a faster and more cost-efficient option for developers and businesses building AI-powered applications. Google is also offering the model at an introductory price that is significantly lower than its predecessor.

Stronger Coding Performance

Google says Gemini 3.7 Flash has been developed to handle software engineering tasks with greater accuracy and consistency.

The model is designed to generate more production-ready code on the first attempt, potentially reducing the number of corrections and additional prompts developers need during the development process.

Its improvements are also aimed at developers working on complex software projects where following detailed instructions and managing multiple steps are important.

Better Web Development

Web development is another area where Google says Gemini 3.7 Flash has made notable progress.

The model can create more functional layouts and feature-complete web applications with fewer prompts. It is also designed to better understand visual references and maintain consistency with a given design.

Google says the model demonstrates improved design fidelity when working from screenshots, images or complete design systems.

According to the company, Gemini 3.7 Flash scored 1,588 on Arena AI's WebDev Arena, outperforming Gemini 3.6 Flash.

Improved Reasoning

Google is also targeting knowledge-intensive professional tasks with the new model.

The company says Gemini 3.7 Flash performs better on complex document processing and business workflow automation, particularly in fields such as finance, law and biosciences.

In the tests cited by Google, the model scored 34.0 per cent compared with 22.0 per cent for Gemini 3.6 Flash on GDP.pdf, while its score on AutomationBench was 30.4 per cent compared with 17.0 per cent for the earlier model.

These results are based on Google's cited evaluations and should not necessarily be treated as a guarantee of performance across every real-world use case.

Focus on AI Agents

Google has also placed greater emphasis on AI agent capabilities with Gemini 3.7 Flash.

The model is designed to follow instructions more closely, plan multi-step tasks and make tool calls more systematically. The company says this more disciplined approach could reduce the need for manual intervention and repeated attempts when AI agents are used in software engineering and other workflows.

For developers, the combination of stronger planning, coding and tool use is aimed at making the model more suitable for production-oriented AI applications.

Introductory Pricing

Google has announced an introductory pricing structure for Gemini 3.7 Flash that will remain in place through the end of 2026.

The model will cost $0.75 per one million input tokens and $3.75 per one million output tokens during the introductory period.

Google is positioning the pricing as an effort to make large-scale deployment of AI agents and production applications more affordable for developers and businesses.

New Safety Measures

The company has also introduced additional safeguards with Gemini 3.7 Flash.

Google says the model includes updated protections against potential misuse in areas involving chemical, biological, radiological and nuclear (CBRN) risks, as well as cybersecurity.

The company said it has strengthened jailbreak protections while continuing to support legitimate applications in areas covered by its broader bioresilience and cybersecurity programmes.

Model Still Has Limitations

Despite the improvements, Gemini 3.7 Flash retains some of the limitations commonly associated with foundation models.

Google notes that the model can still produce inaccurate or fabricated information, while users may occasionally encounter slow responses or timeout issues.

The company also states that the model's knowledge is current through March 2026 for some domains, while certain areas may still depend on information available only through January 2025.

Google's AI Push Continues

The launch of Gemini 3.7 Flash comes as Google continues to expand its AI model lineup and compete aggressively in coding, software development and AI agent applications.

With its focus on lower costs, stronger coding performance, multi-step reasoning and tool use, the new Flash model is aimed particularly at developers and organisations looking to integrate AI into everyday workflows.

However, as with other foundation models, users will still need to verify outputs, particularly when deploying the system for important technical, business or professional tasks.

Google has entered the Bluetooth tracker market with the launch of the Pixel Tag in India, taking on Apple's AirTag with a tracker designed around the Android ecosystem. The device was introduced at Google's Made by Google 2026 event and uses the company's Find Hub network to help users locate lost belongings.

The Pixel Tag can be attached to everyday items such as keys, wallets and bags. When an item goes missing, users can use the Find Hub app to check the tracker's location and, when it is within Bluetooth range, make the device play a sound.

Google's Answer to AirTag

The biggest difference between Pixel Tag and Apple's AirTag is the tracking network they use.

AirTag relies on Apple's Find My network, while Pixel Tag uses Google's Find Hub network. Google says Find Hub can draw on more than one billion Android devices, potentially allowing a Pixel Tag to be located even when it is outside the direct Bluetooth range of its owner.

The tracker can be paired with Android smartphones running Android 9 or later. Users can give their trackers custom names, view their location on a map and share access with up to 10 people.

More Accurate Location Tracking

Google has equipped Pixel Tag with technologies designed to make short-range tracking more precise.

The tracker supports Ultra-Wideband (UWB) as well as Bluetooth Channel Sounding, allowing compatible devices to provide more accurate directional guidance when searching for a misplaced item.

When the tracker is nearby, its built-in speaker can also be activated to help users identify its exact location.

Gemini Integration

Google has also brought its Gemini AI assistant into the Pixel Tag experience.

Users can ask Gemini to help locate or ring their Pixel Tag. This provides an additional way to find a misplaced item without having to manually navigate through the tracking application.

The tracker can also work in the opposite direction. A physical button on the Pixel Tag can be pressed to make a connected Pixel phone ring, even when the phone is set to silent mode.

Alerts for Forgotten Items

The Pixel Tag includes features designed to reduce the chances of users leaving important belongings behind.

Users can receive alerts when they leave a tagged item behind. Google has also included an unknown tracker alert, which can notify users if an unfamiliar tracker appears to be travelling with them.

The feature is intended to provide protection against unwanted or unauthorised tracking and addresses one of the key safety concerns associated with Bluetooth trackers.

Battery Lasts Over a Year

The Pixel Tag uses a replaceable CR2032 coin-cell battery. Google says the battery can last for more than a year, reducing the need for frequent replacements.

The tracker also carries an IP67 rating for dust and water resistance. According to Google, it can withstand immersion in up to one metre of water for up to 30 minutes.

Pixel Tag vs Apple AirTag

The Pixel Tag is positioned as Google's Android-focused alternative to Apple's AirTag.

While both products are designed to locate everyday belongings, their biggest ecosystem difference is the network behind them. AirTag is built around Apple's Find My network, whereas Pixel Tag uses Google's Find Hub and the wider Android device ecosystem.

For Android users, particularly those already using compatible Pixel devices, features such as UWB, Bluetooth Channel Sounding and Gemini integration could make Pixel Tag a useful addition to everyday device tracking.

What Makes Pixel Tag Different?

Google's strategy with Pixel Tag goes beyond simply copying the basic Bluetooth-tracker formula. The company is combining its Android ecosystem, Find Hub network and Gemini AI capabilities with hardware features aimed at improving short-range location accuracy.

The success of the Pixel Tag in India will ultimately depend on factors including its pricing, compatibility, network coverage and how effectively Google's Find Hub network performs in real-world situations.

ChatGPT users in India could soon notice a new element alongside their AI-generated answers — advertisements. Some Indian users have reportedly received emails from OpenAI regarding updates to its Privacy Policy that include information about advertising on ChatGPT.

However, OpenAI has not officially confirmed the launch of ChatGPT ads in India. The company's latest public update says ads have expanded to the UK, Mexico, Brazil, Japan and South Korea, while its current information on advertising availability does not list India as an active ad market. (OpenAI)

The development nevertheless signals that OpenAI is continuing to expand its advertising system internationally.

Privacy Policy Gets an Update

OpenAI has updated its privacy information to explain how advertising works on ChatGPT and how users can manage their advertising experience.

The company has said its advertising model is designed around three key principles — independent answers, conversation privacy and user control.

OpenAI's stated policy is that advertisements should remain separate from ChatGPT's responses and should be clearly identified as sponsored content. (OpenAI)

Which Users Can See Ads?

OpenAI's current advertising model is aimed primarily at users on the Free and Go plans.

According to the company's official information, Plus, Pro, Business, Enterprise and Education users will not see ads. Ads are also not shown to accounts where the user says they are under 18 or where OpenAI predicts that the user is under 18. (OpenAI)

The advertising pilot originally began in the US in February 2026 before being expanded to additional markets. (OpenAI)

How Will ChatGPT Ads Appear?

OpenAI says advertisements appear separately from ChatGPT's answer rather than being incorporated into the AI-generated response.

The company's current ad format places sponsored content below ChatGPT responses. Ads are identified with advertiser information and are visually separated from the organic answer. (OpenAI Help Center)

For example, a user researching a product or comparing services could potentially see a sponsored recommendation below the response.

Will Ads Influence AI Answers?

This is one of the key questions surrounding ChatGPT's advertising model.

OpenAI says ads do not influence the answers ChatGPT provides. The company states that responses are optimised according to what is considered most helpful to the user, while advertisements remain separate and clearly labelled. (OpenAI)

This distinction is particularly important because ChatGPT is increasingly being used not only for general searches but also for product research, comparisons and purchasing decisions.

What About User Conversations?

OpenAI says advertisers do not receive users' private conversations or chat history.

According to the company's advertising principles, conversations remain private from advertisers and user data is not sold to advertisers. Advertisers instead receive aggregated information about advertising performance. (OpenAI)

OpenAI also says users can control advertising personalisation and can clear information used for ads.

Why Is OpenAI Introducing Ads?

Advertising gives OpenAI another potential source of revenue while allowing the company to continue offering a free and lower-cost version of ChatGPT.

OpenAI has previously said that maintaining broad access to ChatGPT requires significant infrastructure and ongoing investment. The company says advertising can help support access to its services without changing how ChatGPT works. (OpenAI Help Center)

The company is also building a broader advertising platform. In May 2026, OpenAI announced additional tools for advertisers, including CPC bidding and expanded campaign measurement. (OpenAI)

What Indian Users Should Know

For Indian users, the key point is that a nationwide ChatGPT advertising launch has not yet been officially announced by OpenAI.

The company is continuing to expand its advertising pilot into new markets, and India could become part of that expansion in the future. OpenAI's presence in India is already significant: the company said in February 2026 that India had more than 100 million weekly ChatGPT users. (OpenAI)

If advertising is eventually introduced for Indian users, Free and Go users are the groups most likely to encounter sponsored content under OpenAI's current model, while the company's listed paid and organisational plans remain ad-free.

For now, reports of emails received by Indian users should therefore be viewed as an indication of potential preparation or policy changes, not confirmation that ChatGPT ads have officially launched across India.

Accenture has introduced a one-time change to its vacation policy, allowing employees to carry forward unused vacation days into the next financial year. The move comes as the company has asked staff to defer taking vacation during August, traditionally the final month of its financial year.

The temporary change is aimed at ensuring higher employee availability during a crucial period for the company, as Accenture works to improve business performance and meet revenue and bookings expectations.

One-Time Vacation Policy Change

Under the revised arrangement, employees who do not use their available vacation entitlement during the current financial year will be able to carry those unused days into the next financial year, which begins on September 1.

The move represents a departure from Accenture's usual vacation arrangements and has been described as a one-time policy adjustment rather than a permanent change to employee leave rules. (The Economic Times)

The company has effectively given employees an option to postpone planned vacation without losing the unused entitlement at the end of the financial year.

Why Has Accenture Changed the Rule?

The policy change comes after Accenture CEO Julie Sweet asked employees to focus on business activity during August.

The company is approaching the end of its financial year, making the final month particularly important for sales, client engagements and revenue generation.

Sweet has urged employees to look for more ways to serve clients and support the company's business objectives during this period. The unusual vacation request has attracted attention because companies generally encourage employees to use their available leave rather than defer it.

Employees Asked to Defer August Vacation

Accenture's request is primarily focused on August, with employees being encouraged to postpone non-essential vacation plans.

The company has introduced the carry-forward provision to ensure that employees who agree to defer their holidays are not disadvantaged by the request.

The arrangement means eligible unused vacation can move into the new financial year rather than being lost at the usual year-end transition. (mint)

What Happens From September 1?

Accenture's new financial year begins on September 1. Under the one-time arrangement, unused vacation covered by the temporary policy will be carried forward into that new financial year.

Employees therefore have the opportunity to take the deferred leave later instead of rushing to use it before the end of August.

However, the latest reports describe the arrangement as a one-time exception, meaning employees should not assume that the same carry-forward provision will automatically apply at the end of future financial years. (The Economic Times)

Accenture's Regular India Leave Policy

Accenture's benefits can vary depending on the employee's location and applicable rules. For employees in India, the company's official careers information says paid time off includes earned leave, sick leave and casual leave, with eligibility based on the state in which the employee works. (Accenture)

This means employees should distinguish between the company's general leave benefits and the temporary global vacation arrangement reported for the current financial-year transition.

Employees should also check their internal Accenture HR or leave-management portal for the specific terms applicable to their employment location and leave balance.

Pressure on Business Performance

The vacation-policy change comes against the backdrop of concerns over Accenture's business outlook, particularly as the company seeks stronger bookings and revenue growth.

The decision to encourage employees to remain available during August highlights the importance of the final month of the company's financial year.

For employees, however, the temporary carry-forward arrangement provides some flexibility. Those who postpone their August vacation will reportedly be able to use the unused entitlement after the new financial year begins.

What Employees Should Know

Can unused vacation be carried forward? Yes, under the one-time arrangement announced for the current year.

When does the new financial year begin? Accenture's new financial year begins on September 1.

Is this a permanent policy change? No. Current reports describe it as a one-time exception.

Why was the change introduced? Accenture is encouraging employees to defer vacation during August as the company focuses on business activity before the financial year closes. (The Economic Times)

Employees should therefore verify their individual eligibility and the exact number of days covered through Accenture's internal HR systems, as leave rules can differ by country, employment arrangement and applicable local regulations.