News from the AI & ML world
DeeperML - #dataprivacy
|
Ellie Ramirez-Camara@Data Phoenix
//
References:
Data Phoenix
, GZERO Media
Google has made several significant announcements regarding its AI and search capabilities. The company has launched an "AI mode" for Search, an experimental feature designed to handle complex queries that would typically require multiple traditional web searches. This new mode leverages AI to provide more advanced reasoning, thinking, and multimodal capabilities, allowing users to ask intricate questions and receive unified, ordinary language responses. Alongside this, Google has expanded access to its AI Overviews, now powered by Gemini 2.0, indicating a broader integration of AI into its search functionalities.
Google has also released Gemma 3, the latest iteration of its open AI models, aimed at improving AI accessibility. Built on the foundation of Gemini 2.0, Gemma 3 is engineered to be lightweight, portable, and adaptable, enabling developers to create AI applications across a wide range of devices. The models are available in various sizes, catering to different hardware and performance needs. Moreover, the Justice Department has ended its attempt to force Google to sell off its stakes in Anthropic, signaling a shift in the legal landscape surrounding Google's AI investments. Recommended read:
References :
@cloud.google.com
//
Google is making significant strides in artificial intelligence, cloud computing, and data protection. They have announced the preview of A4X VMs powered by NVIDIA GB200 NVL72, which substantially enhances training performance for AI models, offering a fourfold increase compared to previous generations. This new system features 72 Blackwell GPUs and 36 Arm-based NVIDIA Grace CPUs, designed to handle the complex compute and memory demands of advanced AI reasoning models.
Google is also expanding data regions coverage to Google Classroom, allowing customers to select specific geographic locations, such as the U.S. or Europe, for their data at rest, to meet organizational or compliance needs. Furthermore, Google has resolved a tax dispute with Italy by paying over 300 million euros, settling a tax audit covering the period from 2015 to 2020. Chrome’s autofill feature is also cited as significantly improving checkout conversion rates for online shoppers and merchants, demonstrating a practical application of Google's technologies. Recommended read:
References :
Jibin Joseph@PCMag Middle East ai
//
DeepSeek AI's R1 model, a reasoning model praised for its detailed thought process, is now available on platforms like AWS and NVIDIA NIM. This increased accessibility allows users to build and scale generative AI applications with minimal infrastructure investment. Benchmarks have also revealed surprising performance metrics, with AMD’s Radeon RX 7900 XTX outperforming the RTX 4090 in certain DeepSeek benchmarks. The rise of DeepSeek has put the spotlight on reasoning models, which break questions down into individual steps, much like humans do.
Concerns surrounding DeepSeek have also emerged. The U.S. government is investigating whether DeepSeek smuggled restricted NVIDIA GPUs via Singapore to bypass export restrictions. A NewsGuard audit found that DeepSeek’s chatbot often advances Chinese government positions in response to prompts about Chinese, Russian, and Iranian false claims. Furthermore, security researchers discovered a "completely open" DeepSeek database that exposed user data and chat histories, raising privacy concerns. These issues have led to proposed legislation, such as the "No DeepSeek on Government Devices Act," reflecting growing worries about data security and potential misuse of the AI model. Recommended read:
References :
David Gerard@Pivot to AI
//
DeepSeek AI is facing increasing scrutiny and controversy due to its capabilities and potential security risks. US lawmakers are pushing for a ban on DeepSeek on government-issued devices, citing concerns that the app transfers user data to a banned state-owned company, China Mobile. This action follows a study that revealed direct links between the app and the Chinese government-owned entity. Security researchers have also discovered hidden code within DeepSeek that transmits user data to China, raising alarms about potential CCP oversight and the compromise of sensitive information.
DeepSeek's capabilities, while impressive, have raised concerns about its potential for misuse. Security researchers found the model doesn't screen out malicious prompts and can provide instructions for harmful activities, including producing chemical weapons and planning terrorist attacks. Despite these concerns, DeepSeek is being used to perform "reasoning" tasks, such as coding, on alternative chips from Groq and Cerebras, with some tasks completed in as little as 1.5 seconds. These advancements challenge traditional assumptions about the resources required for advanced AI, highlighting both the potential and the risks associated with DeepSeek's capabilities. Recommended read:
References :
Jibin Joseph@PCMag Middle East ai
//
The DeepSeek AI model is facing growing scrutiny over its security vulnerabilities and ethical implications, leading to government bans in Australia, South Korea, and Taiwan, as well as for NASA employees in the US. Cisco researchers found DeepSeek fails to screen out malicious prompts and Dario Amodei of Anthropic has expressed concern over its ability to provide bioweapons-related information.
DeepSeek's lack of adequate guardrails has enabled the model to generate instructions on creating chemical weapons, and even planning terrorist attacks. Furthermore, DeepSeek has been accused of misrepresenting its training costs, with SemiAnalysis estimating that the company invested over $500 million in Nvidia GPUs alone, despite export controls. There are claims the US is investigating whether DeepSeek is acquiring these GPUs through gray market sales via Singapore. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |