English हिन्दी
Connect with us

Tech

DeepSeek: China’s open-source AI research lab taking on OpenAI

Published

on

China’s AI research landscape witnessed a significant breakthrough with the unveiling of DeepSeek-R1, an open-source AI model from DeepSeek, a pioneering lab established by Chinese entrepreneur Liang Wenfeng. Positioned as a challenger to global AI leaders like OpenAI, DeepSeek-R1 excels in mathematical reasoning, code generation, and cost efficiency, marking a pivotal shift in the field.

What is DeepSeek?

DeepSeek emerged from Fire-Flyer, the deep-learning division of High-Flyer, a Chinese quantitative hedge fund. Founded in 2015, High-Flyer built its reputation by leveraging AI in financial analytics. In 2023, Liang Wenfeng, High-Flyer’s founder, redirected resources to create DeepSeek, focusing on advancing AI research.

Operating independently from Chinese tech giants like Baidu and Alibaba, DeepSeek prioritizes scientific exploration over immediate profits. Liang noted, “Basic science research rarely offers high returns on investment,” reflecting his commitment to long-term innovation.

DeepSeek-R1: Redefining AI efficiency

DeepSeek-R1, the lab’s flagship model, demonstrates robust reasoning abilities, surpassing several industry benchmarks. Unlike traditional models relying on supervised fine-tuning, DeepSeek-R1-Zero utilized reinforcement learning (RL) for training. The refined DeepSeek-R1 version enhances readability and language consistency, matching OpenAI’s cutting-edge performance in reasoning tasks.

Key innovations, such as multi-head latent attention (MLA) and a mix-of-experts approach, allow DeepSeek-R1 to operate with just a fraction of the computing power needed for comparable models. For instance, its computational needs are reportedly one-tenth of Meta’s Llama 3.1, a breakthrough highlighted by industry analysts.

A new wave of talent

DeepSeek’s team comprises fresh graduates from leading Chinese institutions like Tsinghua and Peking University. Despite limited industry experience, these researchers bring academic rigor and a collaborative spirit to tackle complex AI challenges. Their collective goal is to overcome technological barriers and elevate China’s global AI standing.

Navigating US restrictions

DeepSeek’s advancements come amid U.S. export restrictions on advanced AI chips. The 2022 embargo limited Chinese firms’ access to hardware like Nvidia’s H100 chips, crucial for AI development. Although DeepSeek started with a stockpile of 10,000 H100s, the restrictions pushed the lab to rethink AI architecture and optimize resource efficiency.

Innovative strategies behind DeepSeek

Faced with hardware limitations, DeepSeek adopted efficiency-focused strategies to refine its model architecture:

  • Custom communication schemes: Streamlined data exchanges to reduce memory use.
  • Memory optimization: Minimized field sizes for enhanced efficiency.
  • Mix-of-models approach: Combined smaller models for superior results.

These techniques have set DeepSeek apart as a leader in resource-efficient AI research.

Global impact through open sourcing

DeepSeek’s decision to open-source its models under an MIT license signals a democratization of AI research. By providing access to its advanced models and training techniques, the lab invites global developers to refine and expand its technology. This move not only challenges Western AI dominance but also reinforces DeepSeek’s position as a forward-thinking pioneer in the field.

As DeepSeek continues to push boundaries, it symbolizes a new era of Chinese innovation, blending efficiency with open collaboration to reshape the global AI landscape.

India News

Centre plans stricter compliance rules for VPN providers, may mandate compliance officers

The Centre is exploring stricter compliance requirements for VPN providers, including designated compliance officers, to improve cybercrime investigations and regulatory accountability.

Published

on

The Centre is considering stricter compliance measures for virtual private network (VPN) service providers operating in India as part of its broader efforts to strengthen cybersecurity enforcement and improve cybercrime investigations. Officials are exploring a framework that could require VPN companies to appoint designated compliance officers or authorised representatives to coordinate with enforcement agencies and the Indian Computer Emergency Response Team (CERT-In).

The move comes amid growing concerns within the government over the alleged misuse of VPN services to conceal user identities, bypass law enforcement, and access websites or online platforms that have been blocked in the country.

Government seeks stronger compliance framework

According to government officials, the proposal is intended to ensure that VPN providers establish a clear compliance mechanism in India. Designated officials would be responsible for responding promptly to lawful requests from investigating agencies and CERT-In during cyber incident investigations.

Authorities have stressed that the objective is to improve accountability and facilitate quicker coordination during investigations involving cyber offences rather than monitor ordinary internet users.

Officials also noted that cybercriminals frequently use anonymisation tools, including VPN services, while carrying out ransomware attacks, financial fraud, phishing campaigns, and other online crimes. The proposed framework is aimed at helping investigators trace those involved in such offences through lawful procedures.

Proposal draws from existing IT Rules

The proposed compliance model is similar to the framework introduced under the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021.

Under those rules, significant social media intermediaries are required to appoint a Chief Compliance Officer, a Nodal Contact Person for round-the-clock coordination with law enforcement agencies, and a Resident Grievance Officer. Officials believe adopting a comparable structure for VPN providers would improve accountability, particularly for companies serving Indian users from outside the country.

Existing CERT-In directions remain in focus

The renewed discussions also highlight the cybersecurity directions issued by CERT-In in April 2022.

Under these directions, VPN providers, cloud service providers, virtual private server (VPS) providers, and data centres are required to collect and retain subscriber information for at least five years, even after a customer stops using the service.

The framework requires providers to maintain verified subscriber details, including names, physical addresses, contact numbers, email addresses, IP addresses, the duration of service usage, and the purpose for which the service was obtained. These records must be made available to authorities when sought as part of a lawful cybercrime investigation.

Government officials have maintained that access to subscriber information is limited to lawful investigations and is not intended for indiscriminate surveillance.

VPN industry had opposed data retention rules

When the CERT-In directions were introduced in 2022, several VPN companies and digital rights groups opposed the requirements. They argued that mandatory retention of customer information conflicted with their “no-logs” policies, which are designed to protect user privacy.

Some providers responded by changing their operational models. ExpressVPN, for instance, removed its physical servers from India and began serving Indian users through virtual server locations outside the country. Other VPN companies adopted similar arrangements while continuing to offer services in the Indian market.

Following representations from industry associations and technology companies, the government had extended the compliance deadline from June 27 to September 25, 2022, allowing providers additional time to implement the required changes.

Focus remains on cybercrime investigations

Officials said the latest initiative is intended to place VPN providers under compliance standards comparable to those applicable to other digital intermediaries operating in India. The government believes the proposed measures will strengthen its ability to investigate cybercrimes while further expanding the country’s digital regulatory framework.

Continue Reading

Tech

Google warned millions before Venezuela quake: Can your Android phone save lives during earthquakes?

Google’s Android Earthquake Alerts System sent warnings to users before strong earthquakes hit Venezuela. Here’s how the smartphone-based early warning technology works.

Published

on

Millions of Android users in Venezuela reportedly received earthquake alerts on their smartphones moments before powerful tremors shook the country, renewing global interest in how Google’s earthquake warning technology functions and whether smartphones can help save lives during natural disasters.

The alerts were triggered ahead of the strong shaking caused by the powerful earthquakes that struck Venezuela this week. However, experts have clarified that Google did not predict the earthquakes. Instead, the company’s system detected the initial seismic activity and rapidly issued warnings to people located in potentially affected areas.

How Google’s Android Earthquake Alerts System works

Google’s Android Earthquake Alerts System uses the accelerometers built into billions of Android smartphones. These sensors, typically used to detect device movement and screen orientation, can also identify ground vibrations associated with earthquakes.

When multiple Android devices in the same region detect similar vibrations simultaneously, anonymised data is sent to Google’s servers. Algorithms then analyse the information to estimate the earthquake’s location, magnitude and expected intensity. If the system determines that significant shaking is likely, alerts are automatically sent to users in affected areas.

The technology works by detecting primary seismic waves, or P-waves, which travel faster and are generally less destructive than secondary waves, known as S-waves. Since internet signals travel much faster than seismic waves, alerts can reach users seconds before the strongest shaking arrives.

Why a few seconds can make a difference

Although the warning window is often brief, experts say even a few seconds can be crucial during an earthquake. Such alerts can allow people to move away from hazardous areas, stop dangerous activities, seek cover or prepare for stronger shaking.

During the recent earthquakes in Venezuela, some residents reported receiving alerts on their phones shortly before the tremors intensified, giving them additional time to react.

How to enable earthquake alerts on Android

Google says Android earthquake alerts are available in many countries, including India. To receive these notifications, users need an Android device running Android 5 or later, active internet connectivity and location services enabled.

To check whether earthquake alerts are enabled:

  1. Open Settings on your Android smartphone.
  2. Search for “Earthquake alerts”.
  3. Ensure the feature is turned on.
  4. Keep both location services and mobile data or Wi-Fi enabled.

Google emphasises that the system is designed as a supplemental early warning service and may not detect or alert users about every earthquake. It does not predict earthquakes before they begin but aims to provide precious seconds of advance warning once seismic activity is detected.

Continue Reading

Tech

Meta introduces paid premium features globally for Instagram, Facebook, and WhatsApp

Meta has taken its premium “Plus” subscription plans global for Instagram, Facebook, and WhatsApp, introducing user-experience upgrades while laying out an expensive future roadmap for unified AI and creator tiers.

Published

on

In a significant pivot to diversify revenue streams beyond traditional advertising, tech giant Meta has officially initiated the global rollout of monthly paid subscription plans for its flagship social platforms: Instagram, Facebook, and WhatsApp. The expansion builds upon previous localized testing, bringing a suite of premium user features worldwide.

New premium ‘Plus’ tiers launched

According to statements made by Naomi Gleit, Meta’s Head of Product, the consumer subscriptions are branded under a “Plus” suffix for each respective application. Under the new pricing model, Instagram Plus and Facebook Plus are priced at $3.99 per month, while WhatsApp Plus is available for a slightly lower rate of $2.99 per month. Media reports clarify that these tiers operate entirely independently from the existing “Meta Verified” program, which remains a separate service dedicated solely to account verification and identity protection.

The new subscriptions introduce specific, exclusive upgrades designed for personalization and enhanced user data:

  • Instagram Plus: Grants subscribers the ability to view Stories anonymously without appearing on the viewer list, aggregate insights on how many users rewatched a Story, build unlimited custom audience lists, and extend Story availability beyond the standard 24 hours. Additional perks include customizable bio fonts, exclusive profile pins, unique app icons, and animated “Super Heart” reactions.
  • Facebook Plus: Replicates many of the social expression and profile customization features found in the Instagram tier, alongside the stealth viewing and extended duration tools for Facebook Stories.
  • WhatsApp Plus: Focuses primarily on utility and chat customization, enabling subscribers to change application themes, configure premium ringtones, use exclusive sticker packs, and pin up to 20 chats simultaneously.

Future roadmaps and upcoming AI subscriptions

Beyond consumer-focused social add-ons, Meta revealed it is actively piloting advanced monetization frameworks aimed at creators, business clients, and artificial intelligence enthusiasts. Product lead Naomi Gleit indicated that the company plans to eventually consolidate these upcoming tiers under a unified master brand known as “Meta One”.

Media reports highlight that the AI-centric subscription paths will include a “Meta One Plus” tier at $7.99 per month and a high-end “Meta One Premium” tier at $19.99 per month. While functional capabilities remain identical across both options, the Premium plan allocates significantly higher computing resources, allowing power users to process intensive queries requiring deeper inference and complex logic.

For content creators, Meta is structuring a tiered approach consisting of “Meta One Essential” at $14.99 per month—mirroring legacy verification benefits—and “Meta One Advanced” at $49.99 per month. The high-tier Advanced plan will offer prominent feed visibility, elevated search rankings, and automated follow-invitation mechanisms to maximize creator reach.

This systemic move toward multi-tier subscriptions materializes as Meta experiences heightened infrastructure spending, with internal projections expecting capital expenditure to reach between $125 billion and $145 billion this year due to heavy data center and AI research allocations. Following the formal rollout announcement, public market shares for Meta reacted positively, climbing nearly 3%.

Continue Reading

Trending

© Copyright 2022 APNLIVE.com