---
title: 'AI Archive'
url: 'https://securteksecurity.com/news/archive/ai'
markdown: 'https://securteksecurity.com/news/archive/ai.md'
date: '2026-10-28'
description: 'Rotated-off AI stories, grouped by week. Week of October 21, 2026 Anthropic launches Claude Duet, a proactive agent that comes to you (October 21, 2026) Meta puts Muse at the centre of its strategy: 10 million setups in six weeks, Q3 revenue of $64.6B (October 21, 2026) OpenAI will sell its fronti…'
---

# AI Archive

Rotated-off AI stories, grouped by week.

## Week of October 21, 2026

- [Anthropic launches Claude Duet, a proactive agent that comes to you](https://securteksecurity.com/news/anthropic-launches-claude-duet-proactive-agent) (October 21, 2026)
- [Meta puts Muse at the centre of its strategy: 10 million setups in six weeks, Q3 revenue of $64.6B](https://securteksecurity.com/news/meta-muse-10-million-setups-q3-earnings) (October 21, 2026)
- [OpenAI will sell its frontier models and Codex through Google Cloud](https://securteksecurity.com/news/openai-frontier-models-codex-google-cloud) (October 21, 2026)
- [Nvidia launches the OpenShell Reference Architecture for governing AI agents](https://securteksecurity.com/news/nvidia-openshell-reference-architecture-agent-governance) (October 21, 2026)
- [OpenAI will require approval for model routers and AI gateways from November 20](https://securteksecurity.com/news/openai-require-approval-model-routers-gateways-november-20) (October 21, 2026)
- [Walmart opens Agent Gateway and Agent Profiles for approved shopping agents](https://securteksecurity.com/news/walmart-agent-gateway-agent-profiles-approved-shopping-agents) (October 21, 2026)
- [Recursive releases Flint, a laptop coding model largely built by its Eureka Machine](https://securteksecurity.com/news/recursive-flint-laptop-coding-model-largely-built-by-eureka-machine) (October 21, 2026)
- [Google DeepMind researchers resign over military AI contracts](https://securteksecurity.com/news/google-deepmind-researchers-resign-military-ai-contracts) (October 21, 2026)
- [Dormant agents keep their keys: two reports quantify the agent governance gap](https://securteksecurity.com/news/dormant-agents-keep-credentials-arrakis-illumio-reports) (October 21, 2026)

## Week of October 20, 2026

- [OpenAI says two of its agents weakened the misalignment monitoring meant to catch them](https://securteksecurity.com/news/openai-agents-weakened-misalignment-monitoring) (October 20, 2026)
- [SpaceX Q3: AI segment reaches $7.56B and overtakes Starlink as company turns profitable](https://securteksecurity.com/news/spacex-q3-ai-segment-7-56b-overtakes-starlink) (October 20, 2026)
- [Google and Meta open-source TorchTPU, a native PyTorch backend for TPUs](https://securteksecurity.com/news/google-meta-open-source-torchtpu-pytorch-backend-tpus) (October 20, 2026)
- [DeepSeek open-sources DeepInfer, the production inference engine behind V4.1 Flash](https://securteksecurity.com/news/deepseek-open-sources-deepinfer-production-inference-engine) (October 20, 2026)
- [Meta pauses Muse chat sharing after researchers enumerate 6,800 shared conversations from public DNS records](https://securteksecurity.com/news/meta-pauses-muse-chat-sharing-6800-conversations-public-dns) (October 20, 2026)
- [AWS ships TorchNeuron, an open source PyTorch backend for Trainium](https://securteksecurity.com/news/aws-ships-torchneuron-pytorch-backend-trainium) (October 20, 2026)
- [Claude runs a 64-hour autonomous campaign and produces 17 lab-confirmed scorpion-toxin binders](https://securteksecurity.com/news/claude-64-hour-campaign-17-scorpion-toxin-binders) (October 20, 2026)
- [Claude cuts OpenSSH pre-authentication attack surface by about a fifth; 493 issues flagged across 25 projects](https://securteksecurity.com/news/claude-openssh-attack-surface-493-issues) (October 20, 2026)
- [Mistral ships Voxtral 3 and ElevenLabs ships Eleven Realtime as voice agents become a full stack race](https://securteksecurity.com/news/mistral-voxtral-3-elevenlabs-eleven-realtime-voice-agents-race) (October 20, 2026)

## Week of October 19, 2026

- [DHS orders DeepSeek out of all federal civilian agencies, including tools built on its models](https://securteksecurity.com/news/dhs-orders-deepseek-out-federal-civilian-agencies) (October 19, 2026)
- [UK FCA to build a regulatory framework for agentic finance](https://securteksecurity.com/news/uk-fca-build-regulatory-framework-agentic-finance) (October 19, 2026)
- [EU publishes first harmonised AI Act standard, EN 18286:2026](https://securteksecurity.com/news/eu-publishes-first-harmonised-ai-act-standard-en-18286) (October 19, 2026)
- [GSA LLM contract clause 552.239-7001 takes effect](https://securteksecurity.com/news/gsa-llm-contract-clause-552-239-7001-takes-effect) (October 19, 2026)
- [MiniMax releases M3.1 Flash, and confirms it was the stealth Space Bunny model](https://securteksecurity.com/news/minimax-m31-flash-revealed-space-bunny-stealth-model) (October 19, 2026)
- [Inception launches Mercury 3, its flagship diffusion LLM, at 1,068 tokens per second](https://securteksecurity.com/news/inception-mercury-3-flagship-diffusion-llm) (October 19, 2026)
- [The Information reports DeepSeek will pre-train on Huawei Ascend 950DT by end of 2026](https://securteksecurity.com/news/deepseek-pre-train-huawei-ascend-950dt-information-reports) (October 19, 2026)
- [Santander and NVIDIA: specialised small models cut serving cost in selected experiments](https://securteksecurity.com/news/santander-nvidia-small-models-cut-serving-cost-selected-experiments) (October 19, 2026)
- [fal launches FLUX 3 Max video model, 22 times faster than Seedance 2.5 at the same reported quality](https://securteksecurity.com/news/fal-flux-3-max-video-model-speed) (October 19, 2026)
- [Stripe Link agent purchases rose 38-fold in September, Upstarts Media reports](https://securteksecurity.com/news/stripe-link-agent-purchases-38-fold-september) (October 19, 2026)

## Week of October 12, 2026

- [Character.AI files Chapter 11; Disney is the $20M stalking-horse bidder](https://securteksecurity.com/news/character-ai-chapter-11-disney-stalking-horse-bidder) (October 13, 2026)
- [IMF names an AI correction a top global financial-stability risk](https://securteksecurity.com/news/imf-ai-correction-top-financial-stability-risk) (October 13, 2026)
- [Judge grants Apple expedited discovery against OpenAI](https://securteksecurity.com/news/judge-grants-apple-expedited-discovery-against-openai) (October 13, 2026)
- [OpenAI engineer arrested in alleged conference-room bugging case](https://securteksecurity.com/news/openai-engineer-arrested-conference-room-bugging-case) (October 13, 2026)
- [OpenAI and Anthropic urge UK Parliament to pass binding AI rules](https://securteksecurity.com/news/openai-anthropic-urge-binding-uk-ai-rules-parliament) (October 13, 2026)
- [OpenAI hit with proposed class action over children's data](https://securteksecurity.com/news/openai-class-action-childrens-data-minors) (October 13, 2026)
- [Lawmakers ask the FAA to pause its DC-area AI flight-management tool](https://securteksecurity.com/news/lawmakers-faa-pause-ai-flight-management-tool) (October 13, 2026)
- [Eight states sue OpenAI over ChatGPT and child safety](https://securteksecurity.com/news/eight-states-sue-openai-over-chatgpt-child-safety) (October 14, 2026)
- [EU opens first AI Act enforcement actions](https://securteksecurity.com/news/eu-opens-first-ai-act-enforcement-actions) (October 14, 2026)
- [Baker Lab publishes AI-designed proteases in Nature](https://securteksecurity.com/news/baker-lab-de-novo-ai-designed-proteases-published-in-nature) (October 14, 2026)
- [xAI launches Grok Voice 3.0 for voice agents at unchanged pricing](https://securteksecurity.com/news/xai-grok-voice-3-speech-to-speech-launch) (October 14, 2026)
- [Agent micropayments pass 138 million on Coinbase Base via x402, Chainalysis reports](https://securteksecurity.com/news/x402-agent-micropayments-138-million-base-chainalysis) (October 14, 2026)
- [ASML posts record quarter as AI chip-tool demand outruns supply](https://securteksecurity.com/news/asml-record-quarter-ai-chip-tool-demand-outruns-supply) (October 14, 2026)
- [Anthropic adds midterm safeguards and voting information to Claude](https://securteksecurity.com/news/anthropic-midterm-safeguards-democracy-works-voting-information-claude) (October 14, 2026)
- [Hassabis: AI-agent hacks are "early warning shots"; calls for dynamic certification](https://securteksecurity.com/news/hassabis-ai-agent-hacks-early-warning-shots-certification) (October 14, 2026)
- [Nvidia pledges $250M in compute for UK agentic AI safety research](https://securteksecurity.com/news/nvidia-250m-compute-uk-agentic-ai-safety-research) (October 14, 2026)
- [Gartner: more than $3.5T of AI infrastructure spending could face impairment by 2030](https://securteksecurity.com/news/gartner-3-5t-ai-infrastructure-spending-impairment-2030) (October 14, 2026)
- [FTC subpoenas OpenAI and Anthropic over consumer harms, Axios reports](https://securteksecurity.com/news/ftc-subpoenas-openai-anthropic-consumer-harms-axios-reports) (October 15, 2026)
- [Meta releases Muse Spark Open, its most capable US open-weight model](https://securteksecurity.com/news/meta-muse-spark-open-most-capable-us-open-weight-model) (October 15, 2026)
- [Microsoft completes pre-training of MAI-Base-2, a 2.8 trillion parameter model](https://securteksecurity.com/news/microsoft-mai-base-2-pre-training-complete) (October 15, 2026)
- [DeepMind walkthrough identifies an antimicrobial peptide candidate with Antigravity](https://securteksecurity.com/news/deepmind-antigravity-antimicrobial-peptide-candidate) (October 15, 2026)
- [Trump Super Intelligence Task Force holds its first meeting](https://securteksecurity.com/news/trump-super-intelligence-task-force-first-meeting) (October 15, 2026)
- [Investors trim Anthropic IPO valuation expectations; NYC comptroller urges a pause](https://securteksecurity.com/news/investors-trim-anthropic-ipo-valuation-expectations-nyc-comptroller-pause) (October 15, 2026)
- [FSB finalizes AI sound practices, with stronger human oversight for agents](https://securteksecurity.com/news/fsb-finalizes-ai-sound-practices-stronger-human-oversight-agents) (October 15, 2026)
- [DeepSeek outage lasts nearly five hours; week of API spending refunded as credits](https://securteksecurity.com/news/deepseek-outage-nearly-five-hours-api-credits-refunded) (October 15, 2026)
- [Anthropic grants evaluator METR employee-level access, with publication rights](https://securteksecurity.com/news/anthropic-grants-metr-employee-level-access-publication-rights) (October 15, 2026)
- [Google AI news licensing program passes 400 publications, with CNN, the FT, and Axios joining](https://securteksecurity.com/news/google-ai-news-licensing-program-passes-400-publications) (October 15, 2026)
- [TSMC posts sixth straight record profit on AI demand](https://securteksecurity.com/news/tsmc-record-third-quarter-profit-ai-demand) (October 15, 2026)
- [Apollo Research finds GLM-5.3 reasoned about its own training in 10 of 6,452 rollouts](https://securteksecurity.com/news/glm-53-update-aware-reasoning-apollo-research) (October 16, 2026)
- [Anthropic signs 20 year lease for its first Canadian data centre, in Nova Scotia](https://securteksecurity.com/news/anthropic-signs-20-year-lease-first-canadian-data-centre-nova-scotia) (October 16, 2026)
- [Reflection ships Beam open weights under Apache 2.0](https://securteksecurity.com/news/reflection-beam-open-weights-ship-apache-2) (October 16, 2026)
- [Claude Mythos 5 redesigns quantum circuits, cutting gate counts up to 17x](https://securteksecurity.com/news/claude-mythos-5-quantum-circuits-gate-counts-17x) (October 16, 2026)
- [NIST, CISA and NSA publish key practices for super-intelligence agent incidents](https://securteksecurity.com/news/nist-cisa-nsa-key-practices-super-intelligence-agent-incidents) (October 16, 2026)
- [China proposes export controls on 23 AI technologies, including large-model weights](https://securteksecurity.com/news/china-proposes-export-controls-23-ai-technologies-model-weights) (October 16, 2026)
- [Anthropic quietly pitches its own frontier AI safety bill to lawmakers](https://securteksecurity.com/news/anthropic-pitches-own-frontier-safety-bill-lawmakers) (October 16, 2026)
- [Kling AI launches Kling 4.0 video model globally](https://securteksecurity.com/news/kling-4-0-global-launch-video-model) (October 16, 2026)
- [Anthropic report: junior technical hires fell from 27% to 9% as agent hours hit six times staff hours](https://securteksecurity.com/news/anthropic-report-junior-technical-hires-fell-27-to-9) (October 16, 2026)
- [NVIDIA and AI4Bharat open a 2.8 trillion token Indic pretraining corpus and the recipe that built it](https://securteksecurity.com/news/nvidia-ai4bharat-28-trillion-token-indic-corpus) (October 16, 2026)
- [British Airways names Meta AI glasses in its conditions of carriage](https://securteksecurity.com/news/british-airways-names-meta-ai-glasses-conditions-of-carriage) (October 16, 2026)
- [Tests find ChatGPT ads can target by uploaded email, phone and current prompts](https://securteksecurity.com/news/chatgpt-ads-target-email-phone-current-prompts-404-media-tests) (October 16, 2026)
- [Claude traces NSA-supplied NIST curve seeds to lrand48, closing a 25-year cryptography mystery](https://securteksecurity.com/news/claude-traces-nsa-nist-curve-seeds-lrand48) (October 17, 2026)
- [CoreWeave debt trades at distressed levels as AI financing is repriced](https://securteksecurity.com/news/coreweave-debt-trades-distressed-ai-financing) (October 17, 2026)
- [Major websites are closing their doors to AI agents](https://securteksecurity.com/news/major-websites-block-ai-agents-verge-reports) (October 17, 2026)
- [White House puts the FTC in charge of AI law enforcement coordination](https://securteksecurity.com/news/white-house-ftc-leads-ai-law-enforcement-coordination) (October 17, 2026)
- [81 of 134 models fail a counter-terrorism safety benchmark](https://securteksecurity.com/news/tech-against-terrorism-81-of-134-models-fail-benchmark) (October 17, 2026)
- [DeepSeek ships V4.1-Flash-1017 as open weights and opens its Search API](https://securteksecurity.com/news/deepseek-v41-flash-1017-open-weights-search-api) (October 17, 2026)
- [AI-made midterm ads pass $137 million, and about 70% carry no AI label](https://securteksecurity.com/news/ai-midterm-ads-137-million-70-percent-no-label) (October 17, 2026)
- [Replit makes OpenAI its default; CEO says it stopped working with Anthropic, a claim insiders dispute](https://securteksecurity.com/news/replit-makes-openai-default-anthropic-disputed) (October 17, 2026)
- [Anthropic commits 5% of its shares to a new Public Benefit Fund](https://securteksecurity.com/news/anthropic-commits-5-percent-shares-public-benefit-fund) (October 18, 2026)
- [Microsoft pulls out of UK data center deal with Nscale, Telegraph reports](https://securteksecurity.com/news/microsoft-pulls-out-uk-data-center-deal-nscale-telegraph-reports) (October 18, 2026)
- [World launches ID verification for AI agents, starting with banking](https://securteksecurity.com/news/world-launches-id-verification-ai-agents-banking) (October 18, 2026)
- [Project Panama Asian scale documented; Korean publishers appeal for training data disclosure](https://securteksecurity.com/news/project-panama-asian-book-scale-korean-publishers-appeal) (October 18, 2026)

## Week of October 5, 2026

- [Nadella: treat advanced AI as an insider threat, build an emergency brake](https://securteksecurity.com/news/nadella-treat-advanced-ai-as-an-insider-threat-build-an-emergency-brake) (October 9, 2026)
- [Decision models go mainstream: TypeSafe raises $870M, Microsoft ships Decision-1](https://securteksecurity.com/news/decision-models-go-mainstream-typesafe-raises-870m-microsoft-ships-decision-1) (October 9, 2026)
- [Nvidia in talks to deepen Reflection AI bet or buy it](https://securteksecurity.com/news/nvidia-in-talks-to-deepen-reflection-ai-bet-or-buy-it) (October 10, 2026)
- [Amazon in talks to acquire Decart for around $7B](https://securteksecurity.com/news/amazon-in-talks-to-acquire-decart-for-around-7b) (October 11, 2026)
- [Poll: most voters see AI as a job risk, want regulation](https://securteksecurity.com/news/poll-most-voters-see-ai-as-a-job-risk-want-regulation) (October 11, 2026)

---

## Navigation

- Parent: [News Archive](https://securteksecurity.com/news/archive.md)
- Next: [Breaches Archive](https://securteksecurity.com/news/archive/breaches.md)