Tech Meridian
← ENTITY INDEX
COMPANY · ENTITY #2
Related event timeline, sources and context from the news index.
1Password introduced '1Password for Agents', which lets authorized AI agents verify with a unique digital key and request a user-approved credential fill so the agent can sign into accounts without ever seeing the password. Approvals occur on the user’s device (similar to 2FA); initial supported agents include Meta Muse, ChatGPT Work, Dots, Claude Duet, Grok Bot and Town, with more integrations planned.
Microsoft published a Science paper reporting that its AI orchestration system MAI-DxO correctly diagnosed 83% of complex New England Journal of Medicine cases in a simulation, compared with 36% for individual generalist physicians using reference materials; specialists or physician teams (who typically handle such cases) were not included in the comparison. The study tested multiple base models (best results used OpenAI’s GPT-5 and GPT-5.5), found DxO added most value on complex cases and weaker models, and Microsoft said it will train a Mayo Clinic-backed healthcare foundation model and run real‑world evaluations at Beth Israel Deaconess.
On Meta’s third‑quarter earnings call Mark Zuckerberg said Muse — the personal AI agent launched to the public last month — could become a "large‑scale" network of agents that transact and interact on users' behalf. He reported more than 10 million downloads in the US and Canada, ~25% daily usage among initial downloaders, and that ~25% of users are shopping via Muse with average purchases above $200, while privacy incidents and enterprise hesitancy were flagged as unresolved risks.
Recursive, a startup led by Richard Socher and co‑founded by Jeff Clune, is offering Flint, an open‑weight 40‑billion‑parameter coding model the company says was largely developed and optimized by its recursive AI system called the Eureka Machine. Recursive shared benchmark claims that Flint outperforms peers on coding tasks and has raised $650 million at a $4.65 billion valuation from investors including Google Ventures, Greycroft, Nvidia, and AMD Ventures.
Meta said it will issue a new class of non-voting shares to fund its AI initiatives, potential acquisitions, and employee compensation, with trading scheduled to begin in 2027. The announcement accompanied Q3 results and comments from CFO Susan Li and comes as Meta ramps AI spending (it forecasts $135–$145 billion in capital expenditures this year) and rolls out products like the Muse agent and Muse Wonder model.
Meta and CEO Mark Zuckerberg finalized a Founder Alignment Agreement that links his ability to sell a new class of nonvoting shares to his continued residence where Meta’s headquarters are (effectively California); he can sell 20% immediately to cover potential taxes and the remaining 80% in chunks over eight years if he stays. The move responds to the proposed Proposition 40 one-time billionaire tax and is intended to keep Zuckerberg near Meta Superintelligence Labs and help raise funds for AI data centers without diluting his voting control.
Meta told investors on its Q3 earnings call that Muse has been adopted by more than 10 million users and that roughly 25% of people who set up Muse use it daily. CEO Mark Zuckerberg highlighted shopping as a top activity—over a quarter of engaged shopping users recorded average purchases above $200—cited Plaid usage, described Muse Secure VM and a forthcoming Muse Confidential VM, and said Meta expects to improve compute cost-efficiency by over 10x and scale Muse toward profitability and higher revenue per user.
Walmart plans to let customers connect their own AI agents to its site next month via a new Agent Gateway that uses the open Model Context Protocol. The system includes Agent Profiles, default purchase approval, optional spending limits, and rules to prevent agents from changing payment methods or accessing unnecessary personal shopping history; Walmart says it will block agents that violate those terms.
Security firm Claroc reported that Muse share links use the domain cf.metaaiusercontent.com and include the conversation title in the subdomain, allowing researchers to enumerate about 6,800 shared links via passive DNS and access dozens of conversations containing sensitive personal and business information. Claroc called the phenomenon "GhostShares," said the behavior appears to be by design, and reported contacting Meta without a reply; the disclosure follows other Muse security concerns including a recent macOS zero-day reported by researcher Patrick Wardle.
Google DeepMind announced global availability of SynthID Detector, a tool that checks images, audio, and video for SynthID cryptographic watermarks from partner providers (currently OpenAI, NVIDIA, and Kakao; Apple support is expected soon) via uploads to SynthID.com. The detector returns a yes/no signal but can produce false negatives when content is heavily altered and does not detect media from providers using other watermarking schemes (e.g., Meta or Microsoft).
Google announced TorchTPU, an open-source PyTorch implementation that lets developers run existing PyTorch apps on Google’s TPUs with minimal changes; the move follows Amazon’s recent release of TorchNeuron for Trainium and accompanies efforts to decouple PyTorch from Nvidia-specific code. Google also released a TPU agent suite and updated LiteRT to assist porting and edge deployments.
Meta said it paused Muse’s feature for sharing conversations after security firm Claroc reported finding more than 6,800 Muse share links exposed via public DNS records. Claroc said some discovered pages contained sensitive material (marketing revenue, a loan application, a staff calendar); Meta is reviewing the feature while considering how links can be discovered.
Sensor Tower reports Meta’s AI companion Muse amassed about 10.9 million downloads since its September debut, including roughly 10.2 million in the U.S., becoming one of only a dozen apps to hit 10M U.S. downloads within its first 40 days. Meta has heavily promoted Muse across its ecosystem and via advertising (39% of October ad spend), and Sensor Tower data shows Muse has stronger short-term retention than rivals like ChatGPT, Claude and Grok and reached 2.5 million daily active users on October 16.
The Information reports that OpenAI CEO Sam Altman presented three internal scenarios for governing advanced AI: an IAEA-style international watchdog to inspect systems and monitor compute; a 'Demis Plan' modeled on CERN where leading labs merge or operate under a single umbrella; and an 'Eliezer Plan' in which one firm amasses dominant economic power and could withhold or disable others' compute. The report says OpenAI aims to pick a path within about a year while trying not to fall behind rivals and amid internal and policy pressures.
A technical walkthrough demonstrates using AtScale (C2026.7) as a dimensional semantic layer (modeled in SML) for agentic workflows on Amazon Bedrock AgentCore. The post shows a Strands agent running on AgentCore (with Anthropic Claude Sonnet 4.6 for the demo), how AtScale exposes a single model over SQL/XMLA/MCP, identity and RBAC propagation through AgentCore to AtScale, query pushdown to Amazon Redshift Serverless, and deployment on Amazon EKS.
The US Department of Homeland Security has ordered federal civilian agencies to remove Deepseek AI products from their IT systems and barred new purchases, extending the restriction to affiliated firms and third‑party services described as "built with" Deepseek. The notice cites a 2018 procurement supply‑chain law but gives no deadline or specific security concerns and leaves unclear what "built with" covers (for example, whether AI gateways like OpenRouter are included).
A federal magistrate judge ordered former OpenAI engineer Abdulrahman Alfozan to remain in custody pending trial, citing concerns that his ties to Saudi Arabia—including prosecutors’ claims that the Saudi government offered to help post bail—make him a flight risk. Alfozan was arrested after OpenAI discovered a concealed audio device in an executive conference room; prosecutors say evidence includes alleged WhatsApp contacts with a Saudi royal, possible recruitment for a Saudi AI initiative, and encrypted device data.
Redact.dev found roughly 80 WhatsApp help and FAQ pages whose underlying code contained about 100 hidden text blocks of instructions intended for a Meta large language model (headered HelpContentGalaxyLLMInclude). The exposed guidance—removed after WIRED alerted Meta and attributed to a bug—included “what not to say” rules such as explicit directions not to provide legal advice or cite data‑protection laws.
Anthropic is establishing the Anthropic Public Benefit Fund as a wholly-owned LLC and, subject to stockholder approval, is committing 5% of its shares today with board-discretionary additions of 2% for every 50% increase in company valuation above a baseline. The fund will be used to scale public-benefit activities (e.g., donations, targeted distributions) while none of the equity held by the LLC will be used for lobbying or political giving; Anthropic says roughly 20% of shares will be dedicated to public benefit when combined with prior co-founder and employee pledges. The company also said it will give METR employee-like access to evaluate its models, is in talks with Hugging Face, and has opened an office in Singapore with an MOU with the government to advance safe AI adoption.
Google today made Gemini 4 Argon globally available to developers and enterprises via the API in AI Studio and the Gemini platform, and to Google AI Ultra/Pro subscribers in Antigravity, the Gemini app, and AI Mode in Search. Google positions Argon as a high‑precision frontier model for complex multi‑step tasks across coding, finance, legal, cybersecurity, and science, citing top ranking on the Vals Index, example deployments inside Google (Chrome refactor, ADK optimizations), technical speedups (xTB and Gemma inference via vLLM), and introductory pricing ($2 per million input tokens, $10 per million output tokens, with cached inputs 95% off).
OnlyFans creators told WIRED they are frustrated that Meta’s new AI agent, Muse — an adults-only conversational app — is permitted to engage in explicit sexting with adult users, while many performers face account restrictions and content removals on Meta platforms; creators also raise concerns about deepfakes and whether their data was used to train Muse.
Meta updated its commercial terms to remove a clause that said the company "may use any content from such domains, web pages, or URLs" to develop, improve, or train its products — language critics said implied it could scrape news sites and override robots.txt. Meta spokespeople said the change clarifies the consolidated crawling terms applied only to ads and business products and does not alter how the company uses web crawlers; content owners can still signal preferences via robots.txt.
Replit CEO and co-founder Amjad Masad said onstage at TechCrunch Disrupt that Replit is no longer working with Anthropic, citing cost, partnership and performance reasons and noting a recent OpenAI partnership; however, an anonymous source and a September Replit blog post citing Claude Fable 5 and 5.1 suggest Anthropic products may still be in use.
Multiple popular websites and services are preventing autonomous AI agents — including Meta’s Muse, OpenAI’s Dots, xAI’s Grok, Instinct, and Perplexity — from performing tasks like shopping, booking, and account access. Companies such as Amazon, Yelp, airlines, banks, and Cloudflare’s anti-bot tools are cited as blocking agents or requiring paid data licensing, while some retailers (Walmart, Gap, Sephora, Best Buy, Shopify merchants) have partnered to allow agent access.
Nieman Lab reports that Meta’s new commercial terms, effective October 30, 2024, authorize automated crawling, indexing, and use of content from domains “associated” with business accounts on Facebook, Instagram, WhatsApp and Messenger; Meta may use that content to develop, research, improve and train its AI products. The terms say businesses that do not accept them may lose access to Meta’s products for commercial use, and Nieman Lab noted this could let Meta access publisher content beyond formal licensing deals.
Since its launch last month, Meta's Muse has been downloaded millions of times and reached No. 1 on the App Store; its onboarding and UI encourage users to grant access to contacts, messages, banking details and other sensitive information that can implicate people beyond the account holder. Security and privacy experts (Signal Foundation, EFF) say Muse’s design and dark patterns risk nonconsensual exposure of third-party data, while Meta says access is opt-in and attributes incidents to misconfigured permissions.
Meta published Muse Spark Open, an open-weight mixture-of-experts (MoE) model with ~816 billion parameters (≈42B active per token) that accepts text, images, audio, and video and supports up to a 1,048,576-token context window. Meta says the model is optimized for agentic coding workflows (reading, writing, debugging code and tool calls); checkpoints are available in BF16 (safetensors, ~1.5 TB) and an NVFP4 4-bit format tuned for NVIDIA Blackwell GPUs.
Meta’s coding-optimized Muse Spark Open (Mixture-of-Experts, 816 billion parameters with 42B active per token and a 1 million-token context window) is now available in Amazon SageMaker JumpStart as two variants (Muse-Spark-Open-816B-A42B BF16 and Muse-Spark-Open-816B-A42B-NVFP4 NVFP4). The model’s open weights (released under the Muse Model License) can be deployed to SageMaker AI endpoints in your AWS account, with instance recommendations ml.p6-b300.48xlarge for BF16 and ml.p5e.48xlarge for NVFP4 and support for the model’s agentic reasoning and multimodal inputs.
NVIDIA announced Nemotron 3.5 Super VL, an open 120B-parameter mixture-of-experts multimodal model that activates 12B parameters per token to support agentic workflows combining reasoning, instruction following, tool use, and text, image, document, and video understanding. NVIDIA says it achieves up to 5× higher single‑GPU throughput on agentic tasks versus comparable open models, supports post‑training via NeMo (SFT, LoRA, RL) with resources on Hugging Face and a GitHub usage cookbook, and employs inference optimizations such as LiLiCorr speculative decoding and NVFP4 quantization; partners including Scale AI, Ramp, and AI21 report substantial cost, latency, or accuracy improvements in production workloads.
Rivian CEO RJ Scaringe told TechCrunch Disrupt that the automaker plans to allow third‑party AI agents (examples cited include Anthropic’s Claude, OpenAI’s Dots and Meta’s Muse) to run in its vehicles, using an in‑house agentic framework that routes requests to local or cloud models. He gave no timeline, stressed significant engineering and coordination work is needed, and said agents will not be granted control of vehicle‑level safety systems; Rivian will discuss more at its December “AI & Autonomy Day.”
Allow Google Analytics to measure visits? It stays off until you choose. You can change your choice at any time.