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

Older stories

TechCrunch

OpenAI dismisses three safety researchers after information-handling investigation

OpenAI has dismissed three members of its safety team after an internal investigation found that they mishandled confidential company information, The Wall Street Journal reported. According to an OpenAI spokesperson, the researchers violated company rules by accessing and handling sensitive material outside approved procedures and sharing it with a third-party AI safety organization. The report did not identify the researchers, the organization or the information involved.

The Decoder

Tavus says Griffin avatar was mistaken for a human in 48% of tests

Tavus has presented Griffin, which it describes as the first Human Interaction Model, or HIM. The system is designed for real-time face-to-face video conversations, analyzing speech, facial expressions, vocal tone, gestures and pauses while also generating video responses.

TechCrunch

AI models retain distinctive writing habits despite removing familiar tells

A study by marketing firm Graphite found that leading AI models continue to rely on recognizable words, phrases and sentence patterns, even as they reduce older signals such as em-dashes and “delve.” Researchers identified 13,000 phrases that appeared at least twice as often in AI-generated material as in human writing. Graphite’s chief AI officer Greg Druck said Claude models are moving closer to human word distribution, while GPT models are moving farther away.

The Decoder

Ideogram 4.5 targets more precise image editing

Ideogram has introduced Ideogram 4.5, a model designed to change selected parts of an image while leaving the remaining areas intact. The company says this can help preserve a person’s body shape and the surrounding background when, for example, only their clothing is replaced. Ideogram presents the approach as a response to persistent artifacts that can appear after repeated edits, even as other image models have improved.

MarkTechPost

Cohere launches Embed 5 embedding models for enterprise retrieval

Cohere has released Embed 5, an embedding model family designed for enterprise search, retrieval-augmented generation and agentic retrieval. It includes Embed 5 Pro, aimed at maximum retrieval quality, and Embed 5 Fast, optimized for lower latency and cost during live queries. Both models process text, images and combined text-image inputs, support 100+ languages and accept up to 128K tokens.

The Verge

Judge dismisses antitrust cases over Google’s AI Overviews

A US federal judge has dismissed antitrust lawsuits brought by Chegg and Penske Media Corporation, the parent company of Rolling Stone, against Google. The companies argued that Google’s AI-powered search features were reducing visits to their websites and damaging their revenue.

TechCrunch

Amazon releases open-source Strands Decider 2B model

Amazon Web Services has released Strands Decider 2B, an open-source model designed to choose among predefined options and provide a confidence score for its decision. The model is available now, can run locally, and is intended to offer faster and less expensive decision-making than a full-featured large language model in some automation workflows. OpenAI announced a similar offering during the same week.

TechCrunch

Shopify launches Canvas AI tool for building online stores

Shopify has introduced Canvas, a site-building tool that lets merchants create and modify online stores through conversations with the company’s AI agent, Sidekick. The system shows updates as they are made, allowing users to view the whole storefront or zoom in on particular elements.

Ars Technica

Ataraxos defeats elite human player in long-standing Stratego challenge

A research team from Carnegie Mellon, MIT, New York University, and Stanford University has developed an AI system that can defeat top human Stratego players. Called Ataraxos, it beat Pim Niemeijer, widely regarded as one of the strongest Stratego players ever, by 15 games to one, with four draws. The system was trained using 16 GPUs and a budget of a few thousand dollars.

TechCrunch

Brian Chesky calls for an operating system built for AI agents

Airbnb has introduced AI-powered search in its fall product update, while CEO Brian Chesky says the company is still exploring what the best interface for travel will be. He argues that chatbots are poorly suited to browsing and shopping because they show limited options and require repeated exchanges. At the same time, he says travel planning and discovery are part of the experience that Airbnb should preserve. The company is also studying “multiplayer” AI interfaces that would let families or friends plan trips together.

Hugging Face

Olmo-core 3 targets trillion-parameter MoE training

Olmo-core 3, a new version of the framework used to develop Olmo language models, introduces an open training system for large mixture-of-experts (MoE) models. Its developers say the design is intended to scale into the trillion-parameter range while limiting the computational and communication costs associated with routing data among experts.

The Verge

OpenAI unveils Dots agent in competition with Meta’s Muse

OpenAI introduced Dots, a personal assistant agent, at its annual DevDay conference in 2026. Chief executive Sam Altman presented the system as a “real-deal AI” agent powered by GPT-6 Astra. The launch was framed against competition from Meta, whose Muse AI agent platform has experienced early runaway success.