Sakhanda Wire
NVDA $155.46 -0.44% MSFT $506.24 -2.27% GOOGL $280.20 -1.24% META $618.87 -1.20% AMZN $257.89 +0.05%

Latest news

Page 18 of 254

DeepMind

AlphaGenome 2 predicts DNA effects in specific cell types

AlphaGenome 2 is a new DNA-sequence model designed to predict how genetic variants influence gene activity in particular human cell types and states. Its developers say the system links single-cell biology with human genetics, helping researchers investigate cellular mechanisms associated with disease.

TechCrunch

Databricks updates Genie to handle extended, unsupervised business tasks

Databricks has unveiled an updated version of its Genie AI agent, designed to carry out multi-step assignments that may take substantial time. The company says Genie uses an ontology built from information stored on its platform to understand business context and complete tasks without supervision.

The Verge

Google DeepMind unveils AlphaGenome 2 for studying DNA and disease

Researchers at Google DeepMind have introduced AlphaGenome 2, an artificial intelligence model intended to help explain how changes in DNA influence different types of cells. The work addresses a difficult question in biology: how genetic changes can have different effects in heart muscle, skin and immune cells, even though nearly every cell in the human body contains the same DNA.

The Decoder

Anthropic says AI cut AMD chip migration to three days

Anthropic says it has reduced its dependence on Nvidia by training and running Claude across four chip families: AWS Trainium, Google TPUs, Nvidia GPUs and AMD GPUs. Co-founder and Chief Compute Officer Tom Brown said the company treats capacity from these platforms as one interchangeable pool for research, model development, customer demand and internal use.

Ars Technica

AI Formally Verifies Key Cryptographic Guarantees in Signal

Google DeepMind researchers used agentic artificial intelligence to mathematically prove several key cryptographic guarantees of the Signal protocol. The four-week project was presented as a proof of concept and is notable both for the protocol and for the AI methods involved.

TechCrunch

Plurai raises $13 million to improve the reliability of AI agents

AI agents still struggle with complex tasks without close supervision. In benchmark tests, Claude Fable 5.2 completed 82.9 of 100 tasks and ChatGPT-6 Astra completed 75.6, while Claude Opus 4.8 reached 76.8% accuracy in a simulated corporate environment evaluated by EON.

The Verge

Adobe unveils experimental AI tools for 3D scenes and characters

Adobe has announced a set of experimental tools showcased at its annual Max conference. The projects are aimed at changing how images, AI-generated characters and photo edits can be developed. One tool can turn images into explorable 3D scenes, while another is intended to make AI characters easier to create and use in video content.

The Decoder

US data firms sell RL environments to Chinese AI labs

At least nine US data companies are selling about $500 million worth of AI training data to Chinese laboratories, according to a ChinaTalk investigation. The firms named include Mercor, Surge AI, AfterQuery and Turing. Some of the products are reinforcement-learning environments: simulated settings that train models for tasks such as coding, financial analysis and office work.

The Decoder

DeepL introduces voice model for more expressive real-time translation

DeepL is introducing Aure, a model intended to make real-time translations preserve more of the original speaker’s delivery. The company says it transfers tone, emotion, emphasis and speaking pace into the target language, helping translated speech reflect changes in mood and elements such as sarcasm.

The Decoder

Grok-built apps exposed patient data and confidential company documents

Security firm Claroc found that 35,578 of 37,571 Grok-built web apps it examined were accessible without login, or 94.7 percent. The apps are created in SpaceXSI’s Build mode, where users describe a product and Grok generates the code. Claroc says the publishing screen on its paid test account had Anyone with the link selected by default. Individual subdomains also made app addresses discoverable through passive DNS databases.

The Decoder

Enterprise sales lift OpenAI quarterly revenue to $10 billion

OpenAI increased its third-quarter revenue by about 50% from the previous quarter to $10 billion, according to reports. Revenue was $3.9 billion a year earlier, while second-quarter growth had been 17–18%. Despite the expansion, the company’s adjusted operating loss widened from $10.1 billion to $10.9 billion. The measure excludes stock-based compensation, which reached $2.6 billion. Its adjusted operating margin improved to negative 109%, from negative 150% in the second quarter.

The Decoder

UK regulator urges universities to adopt AI-resistant assessments

England’s higher-education regulator, the Office for Students (OfS), says artificial intelligence has made it close to impossible to assess some student coursework fairly. It is urging universities to increase their use of traditional examinations and in-person presentations, and warns that institutions failing to establish dependable assessment methods could face sanctions.